Download Full Text - Maastricht University Research Portal

STUDIES OF THE AFFECTIVE AND DEVELOPMENTAL
DOMAINS OF PSYCHOPATHOLOGY IN PSYCHOSIS
© Copyright, N. Kaymaz, Maastricht 2011
ISBN 978 90 75579 529
Cover illustration: ‘Mijn hersenspinsels en gedachtekronkels’ by Ziad Haider.
Ziad Haider werd in 1954 geboren in Amara, een stad in het zuiden van Irak, gelegen aan de Tigris, zo’n 200 kilometer ten zuiden van Bagdad. In 1972 ging studeren aan de Academie voor Schone Kunsten in Bagdad. In 1992 ontvluchtte hij Irak. Voor vijf jaar verbleef hij achtereenvolgens in Syrië en Jordanië, totdat hij in 1997 als vluchteling
aan Nederland werd toegewezen. Vanaf zijn komst naar Nederland in 1997 exposeerde hij ondermeer in Belgie,
Finland en Polen.
Het thema van zijn werk is de onzichtbare erfenis. De blijvende erfenis van de ervaring die in mensen besloten ligt.
Ziad schilderde van binnen uit, vertaalde in kleur en beeld dat wat mensen beleven.
Het leven van het individu en een volk. Ziad Haider is in 2006 in Amsterdam overleden.
www.ziadhaider.net
Cover design & printing by Datawyse / Universitaire Pers Maastricht
STUDIES OF THE AFFECTIVE AND DEVELOPMENTAL
DOMAINS OF PSYCHOPATHOLOGY IN PSYCHOSIS
ACADEMIC DISSERTATION
to obtain the degree of Doctor
at Maastricht University,
on the authority of the Rector Magnificus,
Prof. dr. G.P.M.F. Mols,
in accordance
with the decision of the Board of Deans,
to be defended in public on
22nd June 2011 at 14.00 hours
by
Nil Kaymaz
Born 02 July 1972 in Mazgirt, Turkey
Promotor
• Prof. dr. J. van Os
Beoordelingscommissie
• Prof. dr. M. W. de Vries, voorzitter
• Dr. Ph. A. E. G. Delespaul
• Prof. dr. W. Nolen, Universitair Medisch Centrum Groningen
• Dr. F. P. M. L. Peeters
• Prof. dr. D. Wiersma, Universitair Medisch Centrum Groningen
Table of contents
Chapter 1
Introduction
7
Chapter 2
Developmental studies of psychosis
27
Chapter 3
Affective studies of psychosis
83
Chapter 4
Prodromal studies of psychosis
115
Chapter 5
Summary
141
Chapter 6
Future directions
149
Chapter 7
Summary in Dutch, Nederlandse samenvatting
155
Dankwoord
Curriculum vitae
Publications
163
169
171
CHAPTER 1
Introduction
7
Ever since Kraepelin separated the classification of psychotic illnesses into dementia praecox
and manic-depressive insanity, there has been controversy over whether these two disorders
– now referred to as schizophrenia and bipolar disorder1 – are really distinct from each other.
From a clinical perspective, the disorders may be distinguishable, more specifically in their
seldom occurring pure forms, but there are no pathognomonic symptoms to differentiate
them. During an exacerbation of bipolar disorder, many patients have first rank symptoms of
schizophrenia in the form of characteristic delusions and hallucinations, and many patients
with schizophrenia have symptoms of depression and mania. Current psychiatric nosology
classifies schizophrenia and bipolar disorder as two distinct diagnostic categories with presumed independent aetiologies. There are, however models that challenge the traditional
explanations of the aetiology of psychosis. One view is the categorical approach and the
conceptualization of these as two distinct disorders in the classification systems of the American Psychiatric Association: Diagnostic and Statistical Manual of Mental Disorders (DSM) and
the World Health Organization: International Statistical Classification of Diseases and Related
Health Problems (ICD). The categorical approach is highlighted below for the two main disorders we focused on in this thesis: schizophrenia and bipolar disorder. The alternative view is
that the different clinical forms of psychosis represent a continuum, with schizophrenia representing expressions at the most extreme end of the spectrum (Crow TJ, 1986) and schizoaffective disorder embodying the solution for the diagnostic conundrum induced by the considerable overlap between these disorders (Fig. 1). The main assumptions within the continuum theory are: 1) all diagnoses of psychotic disorder share genetic causes and, 2) the disorders show continuity not only with each other, but also with normal mentation in the general
population. Thus, the notion of a continuum in this thesis is used to denote a continuum of
disorders within the clinical population as well as a continuum of expression of psychosis
across clinical and non-clinical populations.
A debate currently persists regards the merit of old diagnostic traditions in psychiatry. It
is becoming clear that the strict separation of diagnostic entities set forth by the DSM and
ICD classification systems has resulted in major missed opportunities to study the causes and
treatments of psychiatric disorders. Most of the symptom domains, whether it is psychosis,
mania, depression, cognitive impairment or negative symptoms, are found across the entire
psychosis spectrum, which includes all non-affective and affective psychotic disorders. An
elegant way of dealing with the discussions of categorical versus dimensional/continuum
theory is by simply adding the dimensional indicators mentioned above to the categorical
system, as it is depicted in Figure 1. In doing so, the basis for a new cross-diagnostic approach
in psychiatry may be laid.
1
The term bipolar disorder replaced name manic-depressive disorder in the American Psychiatric Association’s DSMIII published in 1980.
8
Figure 1 Three hypothetical patients diagnosed using a combination of categorical and dimensional representations. The categorical diagnoses of schizophrenia, bipolar disorder and schizoaffective disorder are accompanied by
the patients’ quantitative scores (connected by red lines) on five main dimensions (psychosis, negative symptoms,
cognitive impairment, depression and mania) of psychopathology (Kaymaz et al., 2009 and Van Os & Kapur, 2009).
In this thesis, we approached affective and non-affective psychotic disorders as entities pertaining to a multidimensional syndrome. Most of the studies in this thesis focused on schizophrenia as an example of non-affective psychotic disorder, and bipolar disorder as an example of affective psychotic disorder. We examined the epidemiological and genetic overlap
between the two constructs by focusing on developmental and affective domains, and by
examining how this is applicable to the ontogenesis of disorder, i.e. in the prodromal stage as
well as in the context of prediction. We also investigated how the symptom dimensions
found in affective and non-affective psychotic disorders co-vary with and affect each other at
the subclinical and clinical level.
General background information on bipolar disorder and
schizophrenia Bipolar Disorder (type I and type II)
Bipolar disorder is a chronic disorder associated with episodic extremes in mood, varying
from depressive episodes to hypomanic (type II) or manic (type I) episodes. Patients may still
experience residual symptoms in the periods between alternations of depressive and manic
episodes, or between depressive and hypomanic episodes2. The longitudinal outcome of
bipolar disorder is defined by recurrent manic and/or depressive mood episodes. These
mood episodes have a severe impact on the lives of patients and their families, including job
performance, personal relationships and responsibilities. Most of the studies have focused
on the syndromal outcome of treatment of patients with bipolar disorder and less on the
2
Mania and depression can co-occur in a mixed state.
9
functional outcome. A study by Strakowski et al. (1998) showed a 61% syndromal recovery
for bipolar disorder 12 months after hospitalization, whereas functional recovery was reported to be 36%. In a recent naturalistic study in bipolar I disorder, these numbers were
87.5% and 53.5%, respectively (Montoya et al., 2010).
Restoration of pre-episode quality of life and level of functioning is of primary importance in the treatment of bipolar disorder. In contrast to schizophrenia, very little is known
about the basic clinical epidemiology of bipolar disorder. Studies on the incidence rates of
bipolar disorder report rates varying from 1.7 to 4.5 for every 100,000 people per year (Lloyd
et al., 2005; Ketter TA, 2010). Studies to date may have underestimated true incidence of the
disorder. Earlier studies showed little age difference between men and women at onset, but
more recent studies, using more strict operational criteria, have tended to show a later onset
in women (Arnold LM, 2003; Ketter, 2010). It is also uncertain whether migrants have a
higher incidence of bipolar disorder as well as of schizophrenia; although, earlier incidence
studies have suggested that this may be so (Der & Bebbington, 1987; Van Os et al., 1996).
The prevalence rates vary from 0.5%-5.5% (Angst J, 1998), indicating a broad range possibly
due to variances in the populations studied, but also due to the operationalisation and
course of the illness (Bebbington & Ramana, 1995) and overlap with other psychiatric disorders (Krishinan, 2005), which all make the diagnosis of bipolar disorder difficult.
Reported rates of bipolar syndromes are highly variable between studies because of age
differences, differences in diagnostic criteria or restriction of sampling to clinical contacts. In
a study of 1395 adolescents fourteen to seventeen years of age, Tijssen et al. (2010) measured DSM-IV hypomanic and manic episodes (in combination), use of mental health care and
five ordinal subcategories representing the underlying continuous score of hypomanic and
manic symptoms3 at baseline and approximately at 1.5, 4 and 10 years later. Incidence rates
(IR) of mania and hypomania, defined both as DSM-IV episodes and as at least one DSM-IV
core symptom were far higher than traditional estimates. In addition, independent of childhood disorders such as attention deficit hyperactivity disorder (ADHD), the risk of developing
hypomanic or manic episodes was very low after reaching twenty-one years of age. Most
individuals with hypomanic and manic episodes were never in care (87% and 62%, respectively) and did not present with comorbid depressive episodes (69% and 60%, respectively).
The probability of mental health care usage increased linearly with the number of symptoms
on the mania symptom scale. The incidence of the bipolar categories, in particular at the
level of clinical morbidity, was strongly associated with previous childhood disorders and
male gender. Therefore, this was the first time a study showed that experiencing hypomanic
or manic symptoms is a common adolescent phenomenon that infrequently predicts the use
of mental health care. The findings suggest that the onset of bipolar disorder can be elucidated by studying the pathway from non-pathological behavioural expression to dysfunction
and the need for care (Regeer et al., 2009). This is similar to findings reported in nonaffective psychosis, which studies have shown also appears as a distribution in the general
population (Van Os, Linscott, Myin-Germeys, Delespaul, & Krabbendam, 2009).
3
Tijssen et al. (2010) used a 'mania symptom scale' based on the CIDI.
10
Schizophrenia
Schizophrenia is a major mental illness that frequently runs a chronic course, despite the
enormous therapeutic advances from the new findings in research, which look not only at
the aetiological findings, but also at the protective factors. Schizophrenia manifests itself
during late adolescence or early adulthood and has profound effects, not only because of the
direct and substantial suffering for the patients but also for their relatives (Sheitman et al.,
1997; De Mamani AG, 2010). The symptoms of schizophrenia can be characterized by making
a distinction between positive symptoms and negative symptoms. Positive symptoms refer to
the abundance or distortion of normal behaviour. These symptoms are hallucinations such as
hearing voices, or delusions, like believing one’s thoughts are controlled by an external force.
Bizarre behaviour caused by delusions and hallucinations and formal thought disorder (incoherence of speech) are also positive symptoms. Negative symptoms refer to the reduction of
normal functions. The negative symptoms consist of flat affect expressed with a monotonous
voice or immobile facial expression, avolition4 or apathy5, alogia6 and anhedonia7 (Andreasen
NC, 1985).
The positive symptoms of schizophrenia are potentially responsive to or reversible with
antipsychotic medicinal treatment. Negative symptoms, however, are usually stable or even
progressive over time and relatively irreversible. The negative symptoms are associated with
poor outcome in the long term. The clinical presentation of the disorder is heterogeneous.
Either the positive symptoms or the negative symptoms predominate in patients. The onset
of negative symptoms tends to occur about five years before the initial psychotic episode,
with onset of positive symptoms much closer to the time of first hospitalization (Häfner et
al., 1999).
The course of the disorder is also variable. Some patients are left with mild exacerbations of symptoms, while severe cases can involve persistent symptoms that result in hospitalization or even permanent hospitalization.
The estimated life risk varies strongly with the presence or absence of certain risk factors, which are called environmental risk factors, and is on average 1% (Mueser & McGurk,
2004). It seems to occur almost one and a half times more in men than in women (Aleman et
al., 2003).
The symptoms of the disorder were fist clustered and described by Kraepelin, who called
the illness dementia praecox in 1919. ‘Dementia’ refers to the clinical and cognitive deterioration that progresses during the course of the disease, and ‘praecox’ refers to the early
onset of the symptoms of the illness. Bleuler, on the other hand, did not agree with Kraepelin
that schizophrenia always had an early onset and that there was an inevitable deterioration
in the course of the illness. In turn, he emphasized some core symptoms of the disorder such
as the loss of goal-directed behaviour, difficulties in thinking straight and flattening of affect.
Bleuler later named the disorder ‘schizophrenia’ in 1923, referring to the disintegration of
personality that could occur in patients. Since then, many attempts have been made to redefine the diagnostic criteria of schizophrenia, resulting in the development of several diagnos4
lack of desire, drive, or motivation to pursue meaningful goals
a state of indifference or the suppression of emotions
6
poverty of speech
7
inability to experience pleasurable emotions
5
11
tic classification systems, such as the American Psychiatric Association’s Diagnostic and Statistical Manual (DSM) and International Classification of Diseases (ICD; World Health Organization, 1992). Today, we are moving from the DSM-IV-TR definition of schizophrenia as a
chronic disorder characterized by psychotic episodes and a decline in social and occupational
functioning toward developing the DSM-V, with the possible inclusion of new diagnostic
entities such as ‘Attenuated Psychotic Symptom Syndrome’. Another major development is
that researchers and clinicians in psychiatry are critical about these classification systems.
They are pose alternative views about categorical systems and introduce dimensional systems. They are also critical about the term schizophrenia (Van Os J, 2009) and opt to redefine
it, as has already been done in Japan in 2006 with ‘Integration Dysregulation Syndrome’. As
mentioned earlier, there is evidence that the symptoms of schizophrenia are also expressed
at subclinical level in the general population and form part of a continuum (Kaymaz & Van
Os, 2010; Linscott & Van Os, 2010; Van Os et al., 2009).
Aetiology of schizophrenia and bipolar disorder; similarities and
dissimilarities
For more than a century, it has been uncertain whether schizophrenia and bipolar disorder
are distinct disease entities with specific genetic and/or environmental causes and neuroanatomical substrates. Several studies have published evidence from their particular perspective, challenging this dichotomy between the categorical view, which relies on the assumption that both diseases are distinct diagnostic entities, and the continuum theory or the
dimensional view, which points out that there is little evidence for any risk factor, symptom,
treatment or course type being specific to any diagnostic category within the functional psychosis (Van Os et al., 1999). It also promotes the idea that different forms of psychosis represent a continuum of disorders with schizophrenia at the most severe end of the spectrum
(Crow TJ, 1986).
From a clinical perspective, there are no sharp symptomatic distinctions between the
two disorders. Indeed, one piece of evidence for the overlap at the clinical level is the invention of the diagnosis schizoaffective disorder, manic type, which is a diagnosis of a psychopathological state that lies in between bipolar disorder and schizophrenia in terms of clinical
symptomatology, showing both psychotic and mania symptoms. Patients with acute mania
often display Schneiderian first rank symptoms, such as auditory hallucinations, broadcasting
of thoughts, controlled thoughts (delusion of control) and delusional perception (WHO,
1973). Patients with depressive psychosis have delusions and hallucinations (Kendell & Gourlay, 1970; Shergill et al., 1999). Many patients with schizophrenia show symptoms of mania
and depression, and in 75% of cases, patients with schizophrenia have depression symptoms
during their first schizophrenic episode (Koreen et al., 1993; Häfner et al., 1999). These early
affective symptoms are often an early expression of an underlying schizophrenia process, or
in some cases, they are viewed as an additional factor that precipitates the onset of psychosis in those who are biologically predisposed (Chadwick & Birchwood, 1994). There are follow-up studies showing that in initially non-psychotic people experiencing auditory hallucinations, those who become depressed have a higher probability of subsequently developing
12
psychotic disorders needing treatment than do those who do not become depressed (Escher
et al., 2002a, 2002b; Krabbendam et al., 2004). Depression precedes the onset of mania in
many cases of bipolar disorder. Other overlapping characteristics between bipolar disorder
and schizophrenia are at a young age at the time of onset of the disorders, which is slightly
earlier in males (Frangou et al., 2002; Kennedy et al., 2002) and they have a frequent occurrence of life events prior to the onset or relapse of illness (Ventura et al., 1989; Bebbington
et al., 1993).
As for the question as to whether or not there is familial co-aggregation between bipolar
disorder and schizophrenia, family studies consistently have shown that the first-degree
relatives of probands who had schizophrenia are at an increased risk for schizophrenia,
schizoaffective disorder and schizotypal personality disorder. Similarly, family studies have
also revealed that first-degree relatives of bipolar probands are at an increased risk of developing bipolar disorder, schizoaffective disorder and unipolar depression (Taylor MA, 1992;
Faridi et al., 2009). However, whether there is a cross-disorder familial risk for bipolar disorder and schizophrenia has been controversial. Inadequate sample sizes in the studies and low
statistical power may have contributed to this lack of evidence. The familial aggregation of
schizophrenia with itself is expected to be greater than its co-aggregation with bipolar disorder, however, most studies have examined the co-aggregation of schizophrenia or bipolar
disorder with itself and few studies have examined their co-aggregation. One of the largest
family studies to date used the Swedish inpatient register and found over 13,000 cases with
schizophrenia and 5000 with bipolar disorder. Cross-disorder incidence ratios were clearly
increased for both full and half siblings, providing evidence for familial co-aggregation between schizophrenia and bipolar disorder (Osby et al., 2001). The study by Lichtenstein et al.
(2009) on the same extended sample shows similar findings. In this study, the question of
whether schizophrenia and bipolar disorder are the clinical outcomes of discrete or shared
causative processes was investigated by assessing the genetic and environmental contributions to liability for schizophrenia, bipolar disorder and their comorbidity by using data from
a multi-generation register. First-degree relatives of probands who had either schizophrenia
or bipolar disorder were at increased risk of these disorders. Half siblings had a significantly
increased risk, but substantially lower than that of the full siblings. When relatives of probands who had bipolar disorder were analysed, increased risk for schizophrenia existed for
all relationships, including adopted children to biological parents with bipolar disorder.
Heritability for schizophrenia and bipolar disorder was 64% and 59%, respectively. Shared
environmental effects were small but substantial for both disorders. The comorbidity between these two disorders was mainly due to additive genetic effects common for both disorders. This study shows, similar to molecular genetic studies, that schizophrenia and bipolar
disorder partly share a common genetic cause. Another meta-analytic study carried out by
Van Snellenberg & De Candia (2009) found familial aggregation between these two disorders,
which supports the continuum model.
Twin studies show that the siblings of index twins with schizophrenia show a similar
prevalence of affective disorder and schizophrenia, while their parents were more likely to
suffer affective disorder than schizophrenia (Shields & Slater, 1975). A study by McGuffin et
al. (1982) showed that in case of identical triplets, two diagnosed with schizophrenia and the
third with bipolar disorder, highlighted how the same genetic makeup could result in diverse
forms of psychosis. The study by Cardno et al. (2002) showed substantial cross diagnosis
13
concordance for monozygotic twins and modest concordance for dizygotic twins. Common
genetic contributions as well as diagnosis-specific genetic contributions to the variance in
liability to schizophrenia and mania were found using model-fitting techniques. However,
diagnosis-specific genetic effects were not found for schizoaffective disorder. The study by
Cardno et al. (2002) shows a significant overlap in the genes contributing to schizophrenia,
schizoaffective disorder and mania.
Both schizophrenia and bipolar disorder have high heritability estimates of around 80%90% (Cardno et al., 2002; McGuffin et al., 2003). There is evidence for an overlap in familial
susceptibility for bipolar disorder and schizophrenia, but there is still a remaining question as
to whether the origin of this overlap is due to genetic overlap, an overlap in causal factors
(environmental factors) or to a genetic and environmental (GxE) interaction. If the overlap is
of genetic origin, molecular genetic studies should report loci in common. Over the past 20
years, numerous genome scans have been carried out to search the susceptibility genes for
both schizophrenia and bipolar disorder: their consistent findings have generated some confusion. Due to power limitations, few studies exceed the genome-wide significance levels and
attempts to replicate a specific finding require even larger samples. In a study by Badner &
Gershon (2002), all the published genome-wide scan studies up until 2002 were identified.
The most significant linkage zone for both schizophrenia and bipolar disorder were the two
zones on chromosome 13q and 22q.
They concluded that these two regions were likely to harbour susceptibility loci common
to these two disorders. A year later, Lewis et al. (2003) published a meta-analysis on the
same subject and found a number of candidate regions for either schizophrenia or bipolar
disorder, but found no overlap of susceptibility regions between the two disorders. These
conflicting results between the two meta-analyses, published only one year apart, may be
due to the differences in statistical methodology and in the selection of datasets. However,
irrespective of the outcome, the genomic regions covered are large and therefore some
overlap in these zones does not necessarily mean that the same genes are involved in both
disorders. However, more studies should be carried out that examine evidence for genetic
overlap between these two disorders. In a study by Fallin et al. (2005), six genes (DPYSL2,
DTNBP1, G30/G72, GRID1, GRM4 and NOS1) showed overlapping suggestive evidence for
association in both disorders.
Unfortunately, many of the molecular genetic findings in schizophrenia and bipolar disorder have not been replicated consistently. The allelic association at the ZNF804A locus is
now one of the most compelling in schizophrenia to date. In a recent study by Williams et al.
(2010), the authors provided additional evidence for association through meta-analysis of a
large data set patients with schizophrenia/schizoaffective disorder (N = 18,945), schizophrenia plus bipolar disorder (N = 21,274) and controls (N = 38,675) and their data supports the
suggestion of overlapping genetic risk between schizophrenia and bipolar disorder. In a recent study by Grozeva et al., (2010), who investigated the overlap in Copy Number Variations
(CNV) between schizophrenia and bipolar disorder, it was found that schizophrenia and bipolar disorder differ with respect to CNV burden in general and association with specific CNVs
in particular. Their data are consistent with the possibility that possession of large, rare deletions may modify the phenotype in those at risk of psychosis. They suggested that those
possessing such events are more likely to be diagnosed with schizophrenia, i.e. psychotic
disorder with a sizeable neurodevelopmental component. Those without them are more
14
likely to be diagnosed with bipolar disorder, i.e. psychotic disorder without a neurodevelopmental component.
Although the genome-wide association studies (GWAS) are widely used in an attempt to
unravel the genetics of psychiatric disorders, in particular schizophrenia, findings are not
conclusive and explain only a fraction of the heritability. The number of studies reporting
new candidate genes for schizophrenia and for brain structures is expanding; however, there
are conflicting results regarding consistency. As linkage analysis and candidate gene hunting
have resulted in a new gene almost on a daily basis, all hope was put on GWAS, with inclusion of many thousands of participants, as a result of collaboration of multiple centres across
the world. Recent papers on genetic findings in schizophrenia, by Ingason et al. (2010) and by
Stefansson et al. (2009) showed associations with several markers; these, however continue
to explain only a fraction of the observed heritability. The influence of CNVs also does not
necessarily conform to classical nosological disease boundaries in that some CNVs increase
risk across a range of neurodevelopmental psychiatric disorders.
As for brain abnormalities being considered neurobiological markers of brain dysfunction
and seen by many as reflecting the genetic basis to these disorders, i.e. representing endophenotypes; these are presumed to be more proximal to the actions of genes than the clinical syndrome. The endophenotype approach can be used to study the variation among candidate neurobiological markers in subjects with increasing genetic risk. The brain abnormalities have been studied more in schizophrenia than in bipolar disorder and are thought to
represent an endophenotype because they are heritable, measurable in both affected and
unaffected subjects, manifest whether or not the illness is active, associated with the illness
in the general population and found more frequently in the unaffected relatives of patients
than controls (Leboyer et al., 1998; Gottesman & Gould, 2003). The research of brain abnormalities in subjects with bipolar disorder and their relatives is less active compared to schizophrenia. Brain imaging studies in schizophrenia have consistently shown reduced whole brain
volume as well as enlarged lateral and third ventricles (Wright et al., 2000). In a study by
Hoge et al. (1999), seven studies were reviewed which had examined brain size in patients
with bipolar disorder and controls and they reported that there were no differences. These
findings were similar to those in the study by McDonald et al. (2004). However, in a recent
meta-analysis brain changes in bipolar disorder (Arnone et al., 2009) were compared to the
brain changes in schizophrenia; the main conclusions were that individuals with bipolar disorder are characterized by significant whole brain and prefrontal lobe reductions and by
enlargement of the lateral ventricles and globus pallidus. These findings did not distinguish
bipolar disorder from schizophrenia, although schizophrenia was characterized by a greater
degree of ventricular enlargement and amygdala volume reduction. The authors concluded
that because no such reduction was found in the two previous meta-analyses by Hoge et al.
(1999) and McDonald et al. (2004), which encompassed 7 and 11 studies, respectively, the
presence of a small effect might require a larger pool of studies to allow detection. The individuals with schizophrenia showed a greater ventricular enlargement compared to those
with bipolar disorder. The similarities in the brain changes, but also the differences (greater
enlargement of the ventricles in schizophrenia) between bipolar disorder and schizophrenia
can be perceived as evidence that they represent the same disorder with a different severity
or, less parsimoniously, as two separate processes with a similar outcome.
15
Objectives and outline of this thesis
In this thesis, several studies are described concerning the developmental and affective aspects of schizophrenia and bipolar disorder, in terms of where they differ and overlap, and
how that potentially overlaps in the prodrome. The genetic and environmental factors,
shared by these two disorders, were investigated. We also examined evidence for the conceptually different models, from an aetiological perspective, of the categorical (point of rarity, developmental specificity) versus the continuum view (mean differences, no specificity)
of bipolar disorder and schizophrenia. In addition, in the search for evidence for aetiological
models explaining psychotic disorders, it is necessary to discuss the diagnostic implications of
differences in developmental and non-developmental risks across the categories. A literature
search of studies concerning these issues was carried out and data was extracted from published work.
One of our publications, the meta-analysis on the maintenance treatment of depression,
may be considered an outlier within the global objectives and themes of this thesis. However, our approach to psychosis is one of a multidimensional syndrome, with depression
being one of the main symptom dimensions (see Figure 1 above). The presence of affective
symptom dimensions increases the probability of early relapse of psychotic disorders and
their presence in the prodromal phase may act as a risk factor for transition to psychosis later
in life (Strakowski et al., 1995; Bechdolf et al., 2002; Lewandowski et al., 2006). Therefore,
the goal of the meta-analysis was to investigate if there is a difference in sensitivity to treatment between first and recurrent episodes of a major depressive disorder. One of our hypotheses was that recurrent depressive patients would have higher relapse rates and would
benefit less from the protective effects of antidepressants. If confirmed, the next question is
if this can be explained by the behavioural sensitisation model. There is evidence (Post RM,
1992) for behavioural sensitisation in the course of affective illness and Post emphasized the
importance of preventing episodes of depression by preventing relapses by prophylactic
treatment to inhibit sensitisation. According to the sensitisation model, a subgroup of patients exists, that becomes more vulnerable or sensitised to affective episode precipitants
with each recurrent episode. For example, in cases of anticipated stresses or imagined losses,
if sufficiently conditioned, the behavioural, physiologic and biochemical alterations usually
associated with an affective episode might be produced. It may also explain how stressinduced mood alterations might become so sensitised that they also occur spontaneously.
The behavioural sensitisation model may not only be applicable to depression but also to
other symptom dimensions, such as psychosis or other psychotic disorders in which depression is a main component, such as bipolar disorder and schizophrenia. Indeed, there is more
and more evidence from the literature that behavioural sensitisation plays an important role
in relapses of psychoses.
Developmental domains
Developmental impairments, during childhood, preceding the onset in schizophrenia in late
adolescence, have been documented extensively for schizophrenia, specifically in the domains of cognition, language, motor performance, social, emotion and behaviour (Jones et
al., 1994; David et al., 1997; Cannon et al., 2002; Blanchard et al., 2010). For affective psychosis, broadly defined, there has been some documentation of developmental impairments,
16
but the effects are not as strong as they are for schizophrenia (David et al., 1997; Van Os et
al., 1997; Crow et al., 1995) and they seem to be present only in early onset cases (Van Os et
al., 1997). Only a few studies have investigated the childhood development preceding bipolar
disorder or mania. In the Dunedin birth cohort (Cannon et al., 2002), one of the larger prospective studies in this area, children who later met criteria for schizophrenia had shown
developmental impairment in all domains tested. This was in contrast to children who later
developed mania, who showed difficulties in social, behavioural, and emotional development, but not in motor, language or cognition. They even showed better childhood motor
performance than the control group (Cannon et al., 2002). It seems that emotional and interpersonal difficulties in childhood reflect a general predisposition to adult psychiatric illnesses, but that early developmental impairments in psychomotor, language and cognitive
function show some specificity to schizophrenia outcomes and are not seen in those who
later develop bipolar disorder. One explanation is that genes responsible for these developmental impairments are associated with schizophrenia (Jones & Murray, 1991), but not with
bipolar disorder, or, that there is a differential occurrence of early environmental insults in
those two disorders (Murray et al., 2004).
In Chapter 2, the neurodevelopmental component, in relation to genetic factors, was investigated in the disorder most strongly associated with developmental impairment,
e.g.schizophrenia. In addition, the degree to which developmental factors discriminate between schizophrenia and bipolar disorder was examined, and to what extent it may be expected that these psychotic phenotypes are distributed categorically or dimensionally.
The first article, ‘Heritability of structural brain traits; an endophenotype approach to
deconstruct schizophrenia’, focuses on the heritability of brain structures, using the endophenotype approach to deconstruct schizophrenia. We examined evidence for genetic influence on aberrant neurodevelopment by conducting a literature search on the heritability of
brain structures in healthy people, in people with schizophrenia and in pedigrees in order to
deconstruct schizophrenia. In addition to this, a search for molecular genetic variation underlying these brain structures was carried out as well.
The second article is called: ‘Murray et al. (2004) revisited: is bipolar disorder identical
to schizophrenia without developmental impairment?’. It is a review of literature focusing
on work published in the past ten years that examines the actual status of the prediction
offered by Murray and colleagues more than five years ago, which said that bipolar disorder
may be largely similar to schizophrenia, but without associated developmental impairment.
This issue is important to investigate, given the upcoming revisions of diagnostic systems in
psychiatry, with the main question of how bipolar disorder and schizophrenia should be discriminated from each other in DSM-V and ICD-11.
In the third article, ‘Extended psychosis phenotype-yes: single continuum – unlikely’,
phenotypic continuity in the sense of continuity with normal mentation, as suggested both
for schizophrenia and bipolar disorder, is examined form the conceptual point of view. Causation predicts the type of distribution of disease; therefore, given the hypothesized multifactorial aetiology for psychotic disorder, we examined the continuum hypothesis of psychopathology, and why people may shift over the continuum from low values, associated with
subclinical psychotic experiences, to higher levels of psychotic disorder.
17
The affective domain
The paradox of emotional dysfunction in psychotic disorder is historic. It was Bleuler who first
argued that problems of affect lie at the heart of schizophrenia and that hallucinations and
delusions are merely accessory and common to many forms of disorders. This was the view
which gave way to the now familiar distinction between affective and non-affective psychosis. However, emotional dysfunction in patients with schizophrenia is common with the core
symptoms and disabilities, develops rapidly and aggressively during the prodrome and early
phase of the illness into a psychotic episode (Harrison et al., 2001). Following the first episode of psychosis, 50% of patients with schizophrenia report post-psychotic depression
(Birchwood et al., 2000) for a period with a high risk for suicide (Westermeyer et al., 1991).
One third of the patients report traumatic reactions, fulfilling the criteria of PTSD (McGorry
et al., 1991), and 50% report a fear of social interaction, i.e. social anxiety disorder (Cosoff et
al., 1998). The most important issue for patients with psychotic disorders is the problem of
developing and maintaining intimate relationships, leaving most of the patients in isolated
and marginalized positions from social networks. The main question is whether these emotional problems are just a part of the psychotic disorder and should be categorized as negative symptoms such as blunted or flattened affect, or if they should be seen as the (childhood) emotional symptoms of (developing) psychotic disorder. The presence of affective
dysregulation may be important, given that the presence of emotional disorders increases
the probability of early relapse, and their presence in the prodromal phase may act as a risk
factor for transition to psychotic disorder (Strakowski et al., 1995; Bechdolf et al., 2002;
Lewandowski et al., 2006). One of the major findings in the study by Debbane et al. (2009) is
that the expression of positive schizotypy during adolescence is modulated by emotional
factors of depression and anxiety. These affective domains of psychosis, whether it is an
affective or a non-affective psychotic disorder, will unfold in a social environment, e.g. the
urban environment, social factors influencing the morbidity and outcome of psychotic disorder.
In Chapter 3, in the first article, ‘Evidence that the urban environment specifically impacts on the psychotic but not the affective dimension of bipolar disorder’, the potential
overlap of environmental factors on the impact of affective and non-affective symptomatology was examined, specifically focusing on urbanicity and what the impact is of urbanicity on
affective and psychotic symptom domains of bipolar disorder. In the first article, the main
effect of urbanicity on the rate of the bipolar phenotype, narrowly and broadly defined, was
examined in relation to specific symptom dimensions.
In the second article, ‘The impact of subclinical psychosis on the transition from subclinical mania to bipolar disorder’, the prevalences of subclinical mania and psychotic symptoms in the general population were examined and compared with their clinical counterparts. In addition, we examined how these subclinical population phenotypes, which are
more prevalent than the clinical counterparts, co-vary with and impact on each other.
In the third article, ‘Evidence that patients with single versus recurrent depressive episodes are differentially sensitive to treatment discontinuation: a meta-analysis of placebocontrolled randomized trials’, we searched for evidence if there is a difference in sensitivity
to treatment in depression, given that this is an affective symptom domain that is pertinent
to psychotic disorder, including schizophrenia. The aim was to examine the issue of sensitisa-
18
tion, in terms of progressively greater sensitivity to environmental stress, using a relapse rate
paradigm following treatment over short or extended periods of time.
Application to the prodrome
In most cases of schizophrenia or bipolar disorder, the onset of disease is not a sudden event.
Before onset, it has been found that in most cases the hospitalization was preceded by a prepsychotic phase in which attenuated or prodromal psychotic symptoms had been present for
about a year, but sometimes as long as five years in the case of non-specific and negative
symptoms (Hafner H, 2001). In this study, the onset of social disabilities also pre-dated the
first admission to hospital by more than a year and pre-dated even the onset of positive
symptoms. As a consequence of these findings, replicated across many studies (Cannon et
al., 2008; McGorry et al., 2009; Valmaggia et al., 2009), the issue of prediction of psychosis
requires further examination, given that: 1) the prodromal phase is present in the majority of
patients with first-episode psychosis (Hafner H, 2001), 2) it is a source of substantial suffering
on the part of patients and their families, 3) deficits occur in the prodromal phase, and 4)
help-seeking frequently occurs in the prodromal phase (Addington et al., 2002). Early detection and early treatment of psychosis may reduce the psychological, social and possibly biological alterations and deficits (Pantelis et al., 2003) that can lead to poor outcome. However,
there is much confusion from different studies about the initial prodrome of psychosis or at
risk mental state as the high-risk studies are based on the attenuated psychotic symptoms in
selected groups of help-seeking individuals. Data are also needed on the prodrome and the
risk of transition to psychotic disorder in representative general population samples.
In Chapter 4, we discuss in the first article, ‘DSM-V and the ‘Psychosis Risk Syndrome’:
Babylonic confusion’, the validity of diagnostic entities in the psychosis spectrum and the
new paradigm shift in clinical psychiatry to introduce a new category called the Psychosis Risk
Syndrome in DSM-V, which now likely be relabelled as Attenuated Psychotic Symptom syndrome. The introduction of this new category is related to 1) important pioneering work on
early intervention in selected high-risk populations and 2) international efforts emphasizing
the major impact of early intervention on the course and outcome of psychotic disorders.
This paper addresses whether the introduction of another new category in psychosis spectrum is required and whether it is based on valid nosological entities.
The second article entitled, ‘The case of the missing evidence: what do subclinical psychosis spectrum experiences predict in unselected representative population samples? A
systematic review enriched with new results’, is a meta-analysis of existing literature, enriched with new results. We review literature on the risk of developing psychotic disorder
given earlier expression of subclinical psychotic experiences in representative, general population samples. We also examine the risk of conversion to non-psychotic, mainly affective
disorders, given presence of subclinical psychotic experiences in the general population and
investigate which symptom factors moderate risk of conversion to a clinical disorder.
In Chapter 5, an English summary of all the studies included in this thesis is given.
In Chapter 6 the findings and conclusions of all studies described in this thesis are discussed,
with recommendations for future research.
In Chapter 7 a Dutch summary of all the studies included in this thesis is given.
19
Background information on the data of the two longitudinal studies
used in the published studies
Introducing the NEMESIS study
In this thesis, we used the data pertaining to the Netherlands Mental Health Survey and Incidence Study (NEMESIS) in several publications (chapter 3, 4). A more detailed outline of the
NEMESIS study is published by Bijl et al., (1998). This is a brief introduction to the NEMESIS
study.
The NEMESIS study is a prospective study of prevalence, incidence and course of psychiatric disorders in a representative sample of non-institutionalized Dutch adults. A total of
7146 men and women eighteen to sixty-four years of age, contacted through a multistage
sample of municipalities and households, were interviewed at home in 1996. The primary
diagnostic interview of the NEMESIS study was the CIDI, which determines the lifetime occurrence of DSM-III-R disorders. The disorders that were included were mood disorders, anxiety
disorders, eating disorders, schizophrenia and other non-affective psychotic disorders and
dependence, and abuse of psychoactive substances. In this thesis, diagnostic data were used
for all disorders with the exception of data regarding eating disorders. Follow-ups of the
NEMESIS sample were scheduled at 12 and 36 months. The NEMESIS study had three measurement occasions: 1996, 1997 and 1999. The net response of the first measurement was
69.7%. All respondents, with or without mental disorders at the time of the initial interview
in 1996, were monitored for the whole duration of the study. There was no attrition that was
differential with regard to mental health.
The objectives of the NEMESIS study are to obtain data on:
1) The prevalence of psychiatric morbidity amongst adults, eighteen to sixty-four years of
age, in terms of subthreshold/subclinical (groups of symptoms that are potentially clinically relevant but that fail to satisfy the DSM criteria for a disorder) or clinical psychiatric
disorders and the co-occurrence of psychological and (mainly chronic) somatic ailments.
Psychiatric disorders are determined with the aid of DSM-III-R classification (APA, 1987).
2) The consequences of mental disorders in terms of care use and care needs, quality of life
and functional impairments.
3) The incidence and course of disorders. By repeating the measurements, the identification of cases and monitoring of the course of existing disorders over time, in relation to
changing life circumstances, was possible. NEMESIS was the first large-scale nationwide
population study that was fully prospective.
4) Determinants of the emergence and the course of mental disorders were identified: in
particular, socio-demographic characteristics, distressing recent and early life events
(e.g. family history), care received, personality and vulnerability traits (e.g. self-esteem,
neuroticism, locus of control) and support from the social environment. Biological and
physical examinations are not provided in the NEMESIS study.
Introducing the EDSP study
In one publication, we also used data from the Early Developmental Stages of Psychopathology Study (EDSP). Detailed descriptions of the EDSP study have been published by Lieb et
20
al., 2000; a brief introduction to the study will be given here. The EDSP study is a prospective
longitudinal study investigating substance use and other mental disorders in a representative
population sample of 3021 subjects, fourteen to twenty-one years of age (birth cohort 19701981) at baseline (T0) living in Munich, Germany. The age range was especially chosen specifically to address the early developmental stages of substance use, abuse and dependence
and other mental disorders. Two follow-up investigations were conducted after the baseline
investigation covering an overall period of three to four years. In the first follow-up moment
(T1) only the younger cohort, fourteen to seventeen years of age at baseline was investigated, in order to be able to focus on the early developmental stages of psychopathology and
substance use. In the final follow-up session (T2), the entire baseline sample was assessed
again. Special design features are the linkage with a family supplement (EDSP-FS), an independent family survey to investigate familial contributions to the development of substance
use and other mental disorders, as well as neurobiological laboratory studies of high-risk
subjects. The response rates in the different waves of investigation ranged from 70.9%-88.0%
and overall response rate was 84.3%. The interviewers in all assessment stages comprised
mainly graduated psychologists. Parental interviewers were in the age range of the targeted
mothers and they were blind to the diagnoses of the respondents. All interviewers received
training for one week for both the computerized as the paper-pencil version of the M-CIDI.
1) The primary goal of the EDSP study is to provide prevalence and incidence estimates of
substance use, abuse and dependence among adolescents and young adults.
2) Secondly, the EDSP aims to study the natural course of early stages of substance use and
substance use disorders (e.g. initiation of use, progression to abuse) over a period of
several years and to identify risk factors that are associated with changes from one stage
to another.
3) Further main goals of the EDSP study are to include the examination of comorbidity of
substance use disorders with each other as well as with mental disorders and the investigation of familial vulnerability factors and mechanisms that may be involved in the onset and course of substance use disorders.
Limitations of (epidemiological) studies
One of the limitations of the NEMESIS data is that, although it was set up to reach a broad
range of populations, it has not been able to reach certain groups, such as people with no
fixed address, those with insufficient proficiency in Dutch and those who are institutionalized
for a prolonged time. It is unclear what the impact is of missing data on these populations.
For example, it is well known that the prevalence of psychiatric disorders is high in the group
of homeless people, about two or three times higher as in the general population (Health
Council, 1995), possibly missing out on an estimated 3%-5% of people with schizophrenia
among the 20,000 homeless people in the Netherlands. The diagnosis of schizophrenia might
not only be limited by this fact, but also because people with a diagnosis of schizophrenia can
be presumed to be less willing or able to take part in an interview. Almost 38% of the people
who, on the reference date, had been in hospital for over one year were diagnosed with
schizophrenia (Mental Health Care Yearbook, 1995/1996). That means the schizophrenia
prevalence found by the NEMESIS study is an absolute minimum. One other limitation con-
21
cerns ethnicity. First generation immigrants were probably not well represented in the NEMESIS study because of language problems (inability to speak Dutch) and/or because of reduced willingness to participate. However, the younger age groups of these ethnic categories
(Turkish and Moroccan origin) were reached by NEMESIS.
The advantage of the NEMESIS study is that a response rate above 60% is satisfactory
and rare, because there are many scientific, non-scientific and commercial surveys in the
Netherlands. The non-response group did not differ significantly from the response group in
terms of psychiatric morbidity; they even had a better mental health. Another strength of the
study is that the instrument (CIDI) used is widely used internationally. This creates possibilities to compare the NEMESIS data with data from abroad. Moreover, the fact that a computerized interviewing procedure has more advantages in terms of banning partial nonresponse, which avoids one major source of errors when data is entered manually. However,
computerized interviews have their disadvantages, one of which was a technical error that
led to missing data of 71 respondents. Another strong feature of the NEMESIS study is the
one-step procedure in which all respondents underwent a full psychiatric interview. This is in
contrast to other population studies in which a two-step approach is used that results in an
indirect estimate. What they entail is a first step when a limited number of people for the
interview are selected, and a second step logistic regression analysis has to be performed to
compute for every score on the screening instrument the probability that the corresponding
subject had a psychiatric disorder. These conversion factors are then applied to the entire
sample to estimate the total prevalence of psychiatric cases (Health Council, 1995), leaving
us with indirect estimates that may contain substantial error. The NEMESIS prevalence rates
are not indirect estimates, and have a greater number of respondents, resulting in more
accurate estimates and narrower confidence intervals than those in other studies.
As for the EDSP study, the strengths of this study are that it is a combined longitudinal
characterization of psychopathology and substance use behaviours and disorders. It is a longitudinal prospective assessment of various risk and protective factors that are involved in
the development of substance use and other mental disorders and conducted in a representative sample of adolescents and in young adults with a characterization of family factors.
The fact that the EDSP study is conducted in a prospective longitudinal fashion, in a representative population sample, is essential for an unbiased characterization of the psychopathology status of the population and the identification of risk factors, in contrast to clinical
samples that are usually influenced by selection bias and in general allow only for retrospective approaches. Therefore, the EDSP study not only provides a description of prevalence and
prospectively observed incidence patterns of substance use and mental disorders in a target
population on various diagnostic levels such as asymptomatic, symptomatic, subthreshold
and full diagnostic level, but it also offers an adequate basis for investigating a variety of
familial, cognitive-behavioural and social risk factors, and in particular, their status as causal
risk factors. The study also enables one to study various public health related consequences
of substance use and other mental disorders prospectively and unaffected by recall bias.
22
References
Addington J, Van Mastrigt S, Hutchinson J, Addington D (2002). Pathways to care: help-seeking behaviour in firstepisode psychosis. Acta Psychiatr Scand, 106(5):358-64.
Aleman A, Kahn RS, Selten JP (2003). Sex differences in the risk of schizophrenia: evidence from meta-analysis. Arch
Gen Psychiatry, 60(6):565-71.
Andreasen NC (1985). Positive vs. negative schizophrenia: a critical evaluation. Schizophr Bull, 11(3):380-9.
Angst J (1998). The emerging epidemiology of hypomania and bipolar II disorder. J Affect Disord, 50(2-3):143-51.
Arnold LM (2003). Gender differences in bipolar disorder. Psychiatr Clin North Am, 26(3):595-620.
Arnone D, Cavanagh J, Gerber D, Lawrie SM, Ebmeier KP, McIntosh AM (2009). Magnetic resonance imaging studies
in bipolar disorder and schizophrenia: meta-analysis. Br J Psychiatry, 195(3):194-201.
Badner JA & Gershon ES (2002). Meta-analysis of whole-genome linkage scans of bipolar disorder and schizophrenia.
Mol Psychiatry, 7(4):405-11.
Bebbington P, Ramana R (1995). The epidemiology of bipolar affective disorder. Soc Psychiatry Psychiatr Epidemiol,
30(6):279-92.
Bebbington P, Wilkins S, Jones P, Foerster A, Murray R, Toone B, Lewis S (1993). Life events and psychosis. Initial
results from the Camberwell Collaborative Psychosis Study. Br J Psychiatry, 162:72-9.
Bechdolf A, Schultze-Lutter F & Klosterkotter, J (2002). Self-experienced vulnerability, prodromal symptoms and
coping strategies preceding schizophrenic and depressive relapses. Eur Psychiatry, 17(7), 384-393.
Bijl RV, van Zessen G, Ravelli A, de Rijk C, Langendoen Y (1998). The Netherlands Mental Health Survey and Incidence
Study (NEMESIS): objectives and design. Soc Psychiatry Psychiatr Epidemiol, 33(12):581-6.
Birchwood M, Iqbal Z, Chadwick P, Trower P (2000). Cognitive approach to depression and suicidal thinking in psychosis. 1. Ontogeny of post-psychotic depression. Br J Psychiatry, 177:516-21.
Blanchard JJ, Kring AM, Horan WP & Gur R (2010). Toward the Next Generation of Negative Symptom Assessments:
The Collaboration to Advance Negative Symptom Assessment in Schizophrenia. Schizophr Bull, Sep 22. [Epub
ahead of print].
Cannon M, Caspi A, Moffitt TE, Harrington H, Taylor A, Murray RM, Poulton R (2002). Evidence for early-childhood,
pan-developmental impairment specific to schizophreniform disorder: results from a longitudinal birth cohort.
Arch Gen Psychiatry. 2002 May;59(5):449-56.
Cannon TD, Cadenhead K, Cornblatt B, Woods SW, Addington J, Walker E, Seidman LJ, Perkins D, Tsuang M,
McGlashan T & Heinssen R (2008). Prediction of psychosis in youth at high clinical risk: a multisite longitudinal
study in North America. Arch Gen Psychiatry, 65(1), 28-37.
Cardno AG, Rijsdijk FV, Sham PC, Murray RM, McGuffin P (2002). A twin study of genetic relationships between
psychotic symptoms. Am J Psychiatry, 159(4):539-45.
Chadwick P, Birchwood M (1994). The omnipotence of voices. A cognitive approach to auditory hallucinations. Br J
Psychiatry, 164(2):190-201.
Cosoff SJ, Häfner RJ (1998). The prevalence of comorbid anxiety in schizophrenia, schizoaffective disorder and bipolar disorder. Aust N Z J Psychiatry, 32(1):67-72.
Crow TJ (1986). The continuum of psychosis and its implication for the structure of the gene. Br J Psychiatry,
149:419-29.
Crow TJ, Done DJ, Sacker A (1995). Childhood precursors of psychosis as clues to its evolutionary origins. Eur Arch
Psychiatry Clin Neurosci, 224: 61-69.
David AS, Malmberg A, Brandt L, Allebeck P, Lewis G (1997). IQ and risk for schizophrenia: a population-based cohort
study. Psychol Med, 27(6):1311-23.
Debbane M, Van der Linden M, Gex-Fabry M & Eliez S (2009). Cognitive and emotional associations to positive
schizotypy during adolescence. J Child Psychol Psychiatry, 50(3), 326-334.
De Mamani AG (2010). Self-conscious emotions, general emotional distress, and expressed emotion in family members of patients with schizophrenia. J Nerv Ment Dis, 198(4), 305-308.
Der G, Bebbington P (1987). Depression in inner London. A register study. Soc Psychiatry, 22(2):73-84.
Escher S, Romme M, Buiks A, Delespaul P, Van Os J (2002a). Independent course of childhood auditory hallucinations: a sequential 3-year follow-up study. Br J Psychiatry, Suppl. 43:s10-8.
Escher S, Romme M, Buiks A, Delespaul P, Van Os J (2002b). Formation of delusional ideation in adolescents hearing
voices: a prospective study. Am J Med Genet, 114(8):913-20.
23
Fallin MD, Lasseter VK, Avramopoulos D, et al. (2005). Bipolar I disorder and schizophrenia: a 440-single-nucleotide
polymorphism screen of 64 candidate genes among Ashkenazi Jewish case-parent trios. Am J Hum Genet, 77(6),
918-936.
Faridi K, Pawliuk N, King S, Joober R & Malla AK (2009). Prevalence of psychotic and non-psychotic disorders in
relatives of patients with a first-episode psychosis. Schizophr Res, 114(1-3), 57-63.
Frangou S, Raymont V, Bettany D (2002). The Maudsley bipolar disorder project. A survey of psychotropic prescribing patterns in bipolar I disorder. Bipolar Disord, 4(6):378-85.
Gottesman II & Gould TD (2003). The endophenotype concept in psychiatry: etymology and strategic intentions. Am
J Psychiatry, 160(4):636-45.
Grozeva D, Kirov G, Ivanov D, et al. (2010). Rare copy number variants: a point of rarity in genetic risk for bipolar
disorder and schizophrenia. Arch Gen Psychiatry, 67(4), 318-327.
Häfner H, van der Heiden W (1999). The course of schizophrenia in the light of modern follow-up studies: the ABC
and WHO studies. Eur Arch Psychiatry Clin Neurosci, 249 Suppl 4:14-26.
Häfner H (2001). Onset and early course as determinants of the further course of schizophrenia. Acta Psychiatr Scan
Suppl. (407):44-8.
Harrison G, Hopper K, Craig T, et al. (2001). Recovery from psychotic illness: a 15- and 25-year international followup study. Br J Psychiatry, 178, 506-517.
Hoge EA, Friedman L, Schulz SC (1999). Meta-analysis of brain size in bipolar disorder. Schizophr Res, 25; 37(2):17781.
Ingason A, Giegling I, Cichon S, et al. (2010). A large replication study and meta-analysis in European samples provides further support for association of AHI1 markers with schizophrenia. Hum Mol Genet, 19(7), 1379-1386.
Jones P & Murray RM (1991). The genetics of schizophrenia is the genetics of neurodevelopment. Br J Psychiatry,
158:615-23.
Jones P, Rodgers B, Murray R, Marmot M (1994). Child development risk factors for adult schizophrenia in the British
1946 birth cohort. Lancet, 19; 344(8934):1398-402.
Kaymaz N, Van Os J (2009). Murray et al. (2004) revisited: is bipolar disorder identical to schizophrenia without
developmental impairment? Acta Psychiatr Scand. Oct;120(4):249-52.
Kaymaz, N., & van Os, J. (2010). Extended psychosis phenotype - yes: single continuum - unlikely. Psychological
Medicine, 40(12), 1963-1966.
Kendell RE, Gourlay J (1970). The clinical distinction between the affective psychoses and schizophrenia. Br J Psychiatry, 117(538): 261-6.
Kennedy BL, Dhaliwal N, Pedley L, Sahner C, Greenberg R, Manshadi MS (2002). Post-Traumatic Stress Disorder in
subjects with schizophrenia and bipolar disorder. J Ky Med Assoc, 100(9): 395-9.
Ketter TA (2010). Diagnostic features, prevalence, and impact of bipolar disorder. J Clin Psychiatry. Jun;71(6):e14.
Review.
Koreen AR, Siris SG, Chakos M, Alvir J, Mayerhoff D, Lieberman J (1993). Depression in first-episode schizophrenia.
Am J Psychiatry, 150(11):1643-8.
Krabbendam L, Myin-Germeys I, De Graaf R, Vollebergh W, Nolen WA, Iedema J, Van Os J (2004). Dimensions of
depression, mania and psychosis in the general population. Psychol Med, 34(7):1177-86.
Krishnan KR (2005). Psychiatric and medical comorbidities of bipolar disorder. Psychosom Med, 67(1):1-8.
Leboyer M, Bellivier F, Nosten-Bertrand M, Jouvent R, Pauls D, Mallet J (1998). Psychiatric genetics: search for phenotypes. Trends Neurosci, 21(3):102-5.
Lewandowski KE, Barrantes-Vidal N, Nelson-Gray RO, Clancy C, Kepley HO, Kwapil TR (2006). Anxiety and depression
symptoms in psychometrically identified schizotypy. Schizophr Res. Apr;83(2-3):225-35.
Lewis CM, Levinson DF, Wise LH, et al. (2003). Genome scan meta-analysis of schizophrenia and bipolar disorder,
part II: Schizophrenia. Am J Hum Genet, 73(1):34-48.
Lichtenstein P, Yip BH, Björk C, Pawitan Y, Cannon TD, Sullivan PF, Hultman CM (2009). Common genetic determinants of schizophrenia and bipolar disorder in Swedish families: a population-based study. Lancet. Jan
17;373(9659):234-9.
Lieb R, Isensee B, von Sydow K, Wittchen HU (2000). The Early Developmental Stages of Psychopathology Study
(EDSP): a methodological update. Eur Addict Res, 6(4):170-82.
Linscott RJ & Van Os J (2010). Systematic reviews of categorical versus continuum models in psychosis: evidence for
discontinuous subpopulations underlying a psychometric continuum. Implications for DSM-V, DSM-VI, and DSMVII. Annu Rev Clin Psychol, 6, 391-419.
24
Lloyd T, Kennedy N, Fearon P, Kirkbride J, Mallett R, Leff J, Holloway J, Harrison G, Dazzan P, Morgan K, Murray RM,
Jones PB; AESOP study team (2005). Incidence of bipolar affective disorder in three UK cities: results from the
AESOP study. Br J Psychiatry, 186:126-31.
McDonald C, Bullmore ET, Sham PC, Chitnis X, Wickham H, Bramon E, Murray RM (2004). Association of genetic risks
for schizophrenia and bipolar disorder with specific and generic brain structural endophenotypes. Arch Gen Psychiatry, 61(10):974-84.
McGorry PD, Chanen A, McCarthy E, Van Riel R, McKenzie D, Singh BS (1991). Post-traumatic stress disorder following recent-onset psychosis. An unrecognized post-psychotic syndrome. J Nerv Ment Dis, 179(5):253-8.
McGorry PD, Yung AR, Pantelis C, Hickie IB (2009) . A clinical trials agenda for testing interventions in earlier stages
of psychotic disorders. Med J Aust, 190(4 Suppl):S33-6.
McGuffin P, Reveley A, Holland A. (1982). Identical triplets: non-identical psychosis. Br J Psychiatry, 140:1-6.
McGuffin P, Rijsdijk F, Andrew M, Sham P, Katz R, Cardno A (2003). The heritability of bipolar affective disorder and
the genetic relationship to unipolar depression. Arch Gen Psychiatry, 60(5):497-502.
Montoya A, Tohen M, Vieta E, Casillas M, Chacón F, Polavieja P, Gilaberte I (2010). Functioning and symptomatic
outcomes in patients with bipolar I disorder in syndromal remission: a 1-year, prospective, observational cohort
study.J Affect Disord. Dec;127(1-3):50-7.
Mueser KT, McGurk SR (2004). Schizophrenia. Lancet, 19;363(9426):2063-72.
Murray RM, Sham P, Van Os J, Zanelli J, Cannon M, McDonald C (2004). A developmental model for similarities and
dissimilarities between schizophrenia and bipolar disorder. Schizophr Res, 71(2-3):405-16.
Osby U, Hammar N, Brandt L, Wicks S, Thinsz Z, Ekbom A, Sparén P (2001). Time trends in first admissions for schizophrenia and paranoid psychosis in Stockholm County, Sweden. Schizophr Res, 47(2-3):247-54.
Pantelis C, Yücel M, Wood SJ, McGorry PD, Velakoulis D (2003). Early and late neurodevelopmental disturbances in
schizophrenia and their functional consequences. Aust N Z J Psychiatry, 37(4):399-406.
Post RM (1992). Transduction of Psychosocial Stress Into the Neurobiology of Recurrent Affective Disorder. Am J
Psychiatry 1992; 149:999-1010.
Regeer E J, Krabbendam L, De Graaf R, Have MT, Nolen WA & Van Os, J (2009). Berkson's bias and the mood dimensions of bipolar disorder. Int J Methods Psychiatr Res, 18(4), 279-286.
Sheitman BB, Lee H, Strous R, Lieberman JA (1997). The evaluation and treatment of first-episode psychosis. Schizophr Bull, 23(4):653-61. Erratum in: Schizophr Bull 2006, 32(2):401.
Shergill SS, Robertson MM, Stein G, Bernadt M, Katona CL (1999). Outcome in refractory depression. J Affect Disord,
54(3):287-94.
Shields J & Slater E (1975). Genetic aspects of schizophrenia. Br J Psychiatry, Spec No 9:32-40.
Stefansson H, Ophoff RA, Steinberg S, et al. (2009). Common variants conferring risk of schizophrenia. Nature,
460(7256):744-7.
Strakowski SM, Keck PE Jr, McElroy SL, West SA, Sax KW, Hawkins JM, Kmetz GF, Upadhyaya VH, Tugrul KC, Bourne
ML (1998). Twelve-month outcome after a first hospitalization for affective psychosis. Arch Gen Psychiatry.
Jan;55(1):49-55.
Strakowski SM, Keck PE Jr, McElroy SL, Lonczak HS, West SA (1995). Chronology of comorbid and principal syndromes in first-episode psychosis. Compr Psychiatry, 36(2):106-12.
Taylor MA (1992). Are schizophrenia and affective disorder related? A selective literature review. Am J Psychiatry,
149(1):22-32.
Tijssen MJ, Van Os J, Wittchen HU, Lieb R, Beesdo K, Mengelers R, et al. (2010). Evidence that bipolar disorder is the
poor outcome fraction of a common developmental phenotype: an 8-year cohort study in young people. Psychol Med, 40(2), 289-299.
Valmaggia LR, McCrone P, Knapp M, Woolley JB, Broome MR, Tabraham P, Johns LC, Prescott C, Bramon E, Lappin J,
Power P, McGuire PK (2009). Economic impact of early intervention in people at high risk of psychosis. Psychol
Med, 39(10):1617-26.
Van Os J (2009). 'Salience syndrome' replaces 'schizophrenia' in DSM-V and ICD-11: psychiatry's evidence-based
entry into the 21st century? Acta Psychiatr Scand, 120(5):363-72.
Van Os J & Kapur S (2009). Schizophrenia. Lancet, 374(9690):635-45.
Van Os J, Linscott RJ, Myin-Germeys I, Delespaul P & Krabbendam L (2009). A systematic review and meta-analysis of
the psychosis continuum: evidence for a psychosis proneness-persistence-impairment model of psychotic disorder. Psychol Med, 39(2), 179-195.
Van Os J, Jones P, Lewis G, Wadsworth M, Murray R (1997). Developmental precursors of affective illness in a
general population birth cohort. Arch Gen Psychiatry, 54(7):625-31.
25
Van Os J, Takei N, Castle DJ, Wessely S, Der G, MacDonald AM, Murray RM (1996). The incidence of mania: time
trends in relation to gender and ethnicity. Soc Psychiatry Psychiatr Epidemiol, 31(3-4):129-36.
Van Os J, Verdoux H, Maurice-Tison S, Gay B, Liraud F, Salamon R, Bourgeois M (1999). Self-reported psychosis-like
symptoms and the continuum of psychosis. Soc Psychiatry Psychiatr Epidemiol, 34(9):459-63.
Van Snellenberg JX & de Candia T (2009). Meta-analytic evidence for familial co-aggregation of schizophrenia and
bipolar disorder. Arch Gen Psychiatry, 66(7), 748-755.
Ventura J, Nuechterlein KH, Lukoff D, Hardesty JP (1989). A prospective study of stressful life events and schizophrenic relapse. J Abnorm Psychol, 98(4):407-11.
Westermeyer JF, Harrow M, Marengo JT (1991). Risk for suicide in schizophrenia and other psychotic and nonpsychotic disorders. J Nerv Ment Dis, 179(5):259-66.
Wright IC, Rabe-Hesketh S, Woodruff PW, David AS, Murray RM, Bullmore ET (2000). Meta-analysis of regional brain
volumes in schizophrenia. Am J Psychiatry, 157(1):16-25.
26
CHAPTER 2
Developmental studies of psychosis
27
HERITABILITY OF STRUCTURAL BRAIN TRAITS:
AN ENDOPHENOTYPE APPROACH TO
DECONSTRUCT SCHIZOPHRENIA
Nil Kaymaz*,y and J. van Os*,z
*Department of Psychiatry and Psychology, University of Maastricht,
P.O. Box 616 (DRT10) Maastricht, The Netherlands
y
Mediant GGZ/Mental Health Care, Postbus 775, 7500 AT Enschede, The Netherlands
z
Division of Psychological Medicine, Institute of Psychiatry, London, United Kingdom
I. Introduction
A. Endophenotypic Approach to Studying Schizophrenia
B. Neuroanatomic Measures as Endophenotype
C. Heritability Estimates to Determine Contribution of Genetic
Factors to Structural Brain Traits
D. Nonhuman Primate Model for Studying Brain Neurostructural Studies
II. Heritability of Brain Structure Phenotypes
A. Heritability of Brain Structures in Healthy Persons
B. Heritability of Brain Structures in Patients with Schizophrenia Compared
to Healthy Controls
C. Heritability of Brain Structures in Nonhuman Primates (Monkeys)
III. Genes for Brain Structures in Healthy and Persons with Schizophrenia
A. Genes for Brain Structures in Healthy Persons
B. Genes for Brain Structures in Patients with Schizophrenia
IV. Limitations and Clinical Relevance of the Studies
A. Limitations of Neuroimaging Studies
B. Limitations of Twin Studies
C. Limitations of Nonhuman Primate Studies
D. Clinical Relevance of the Heritability Estimations
E. Implications of These of Brain Volumes as Endophenotypes
for Genetics of Psychiatric Disorders
References
Structural brain phenotypes are quantitative traits showing considerable
variation in human populations. Quantitative structural brain abnormalities are
also repeatedly reported in patients with psychiatric disorders such as schizophrenia. Studying the genetic and environmental causes of these differences might
therefore highlight biological mechanisms underlying neuroanatomical phenotypes and directly result in the identification of risk factors for schizophrenia.
Heritability estimates indicate a strong genetic component contributing to neuroanatomical phenotypes. Brain structure volumes have substantial heritability
INTERNATIONAL REVIEW OF
NEUROBIOLOGY, VOL. 89
DOI: 10.1016/S0074-7742(09)89005-3
28
85
Copyright 2009, Elsevier Inc.
All rights reserved.
0074-7742/09 $35.00
KAYMAZ AND VAN OS
rates ranging from high (70–95%) for total brain volume, cerebellar, gray and white
matter, and corpus callosum, to moderate (40–70%) determined for the hippocampus, the four lobes (frontal, temporal, occipital, and parietal lobe), temporal horn
volume, brain parenchyma, white matter hyperintensity, and planum temporal
asymmetry. Middle structures of the brain show high heritability scores for the
deeper structures (ontogenetically earlier formed) and moderate heritability scores
for the surface structures. Structures formed earlier in development show consistently higher heritability rates than brain structures formed later in development,
for example, surface structures, which seem to be influenced by environmental
factors. Even higher heritability reaching 0.99 for total brain volume are estimated
in nonhuman primate (NHP) models employing inbred extended pedigree and
highly uniform rearing conditions, reducing the effects of environmental factors.
Applying highly heritable structural brain phenotypes may serve as an endophenotype for gene mapping studies and lead to identification of genes that are
involved in the regulation of human brain volume and the biological mechanisms
involved in the causal mechanisms of psychiatric disorders.
I. Introduction
A. ENDOPHENOTYPIC APPROACH TO STUDYING SCHIZOPHRENIA
Schizophrenia is a highly heritable disorder (Cannon et al., 1998; Cardno and
Gottesman, 2000; Sullivan et al., 2003) characterized by delusions, hallucinations,
disorganized speech (frequent derailment or incoherence), grossly disorganized or
catatonic behavior, and negative symptoms, that is, affective flattening, alogia
(poverty of speech) and avolition (general lack of desire, interest or motivation to
pursue meaningful goals) (DSM-IV-Tr, 2000). It is believed that this complex
phenotype of schizophrenia is a result of underlying complex genetic architecture
involving interactions between multiple loci and environmental factors (van Os
et al., 2008). Another source of difficulty is the reliability of phenotypic assessments. Schizophrenia diagnostics employ a patient’s self-reports rather than
biochemical, electrophysiological, or other reliably measurable biomarkers.
The validity of schizophrenia as a diagnostic entity has therefore been criticized
by a number of scientists and clinicians as lacking in scientific validity and
diagnostic reliability (Bentall et al., 2004; Boyle, 2002; Kendell and Jablensky,
2003; Linscott et al., 2009; van Os, 2009; van Os and Murray, 2008). Due to
the diagnostic heterogeneity and the genetic and phenotypic complexity of
29
HERITABILITY OF STRUCTURAL BRAIN TRAITS
schizophrenia, as well as other complex brain-related disorders, identification of a
genetic basis of this disease is a very challenging task.
The endophenotypic approach is one of several strategies that have been
adopted to deal with the complexities of these disorders and identify underlying
genes (Gottesman and Gould, 2003). The principle of this approach is to reduce
complex phenotypes into components (neurophysiological, biochemical, endocrine,
cognitive, neuropsychological, or neuroanatomical phenotypes) that could be reliably quantified and are under close regulation by genetic variation. The endophenotype (or intermediate phenotype) is therefore a kind of biomarker that, unlike
disease diagnosis, has a clearer underlying biological and genetic basis. They are
expected to be genetically correlated with disease liability, and can be measured in
both affected and unaffected individuals. Since quantitative endophenotypes are
generally closer to the action of the gene and are measurable in both unaffected and
controls, they exhibit higher genetic signal-to-noise ratios and are anticipated to
provide a greater power to localize and identify disease-related quantitative trait loci
(QTLs) than affection status alone (Blangero et al., 2003).
Human studies often use endophenotypes that are related to evolutionary
conserved traits and therefore can be extended to animal models.
B. NEUROANATOMIC MEASURES AS ENDOPHENOTYPE
Volumes of various brain structures are quantitative phenotypes that can be
reliably measured on MRI scans using either automated (voxel-based morphometry method, VBM; deformation-based morphometry; surface-based morphometry)
or manual tracing (region and volume of interest, ROI and VOI method)
(Ashburner and Friston, 2000; Good et al., 2001a; Thompson and Toga, 1997;
Thompson et al., 1996a,b). Brain structure measurements show considerable
interindividual variation in general populations and differences between healthy
and affected persons, and are therefore considered promising candidate endophenotypes that might facilitate investigations of a genetic basis into the natural
variation of brain-related traits in populations and for psychiatric disorders such
as schizophrenia. When measuring brain structure and the heritability of these
structures, normal neuroanatomical variations in the human brain should be
taken into account (Allen et al., 2002). These neuroanatomical variations are
under the influence of variables such as gender (Cosgrove et al., 2007; Giedd,
2008) and age (Carne et al., 2006; Good et al., 2001b), or other components of
genetic variety such as height (Kappelman, 1993). Intelligence seems to be related
to brain volume as well. In humans, brain volume is a quantitative trait with high
heritability (Posthuma et al., 2002b; Thompson et al., 2001, 2002). Previous studies
have shown that brain volume is also correlated with general intelligence, working
memory, perceptual organization, and processing speed, which are also highly
30
KAYMAZ AND VAN OS
heritable (Posthuma et al., 2003; Roth and Dicke, 2005). In recent years, it has
been discovered that the structure of the adult human brain changes when new
cognitive or motor skills, including vocabulary, are learned (Lee et al., 2007), and
that this structural neuroplasticity (increased gray matter volume), after 3 months
of training a visual motor skill, seems to last for at least 3 months without further
practicing (Driemeyer et al., 2008; Ilg et al., 2008). Another example is the impact
of handedness, which has a significant genetic component (Annett, 1974;
McManus and Bryden, 1991) and is strongly related with cerebral asymmetry
(Alexander and Annett, 1996; Geschwind and Galaburda, 1985). To what extend
the cerebral asymmetry is also heritable remains unclear (Geschwind et al., 2002).
The neuroanatomical phenotype and the heritability of brain structures varies
among individuals with a neuropsychiatric disorder, meaning that the influence
the disease state has on the heritability is such that the morphology of brain
structures can be altered by the disease state and these morphologically altered
brain structures can be inherited. For example, there are studies showing small
hippocampal volumes in adult humans with recurrent major depression (Bremner
et al., 2000; Sheline et al., 1999) and posttraumatic stress disorder (Bremner et al.,
1995). Hippocampal morphology is also altered by stress in carefully controlled
studies with rodents (Gould and Tanapat, 1999; Magarinos et al., 1995; Sapolsky,
2000). These smaller hippocampi have been taken as evidence that stress-related
disorders induce hippocampal volume loss (Sapolsky, 2000; Sheline et al., 1999).
One possible pathway by which altered brain morphology changes might be
inherited is that small hippocampal volumes are inherited and act as predisposing
factors toward the development of psychiatric disorders that are triggered by
stress (Gurvits et al., 1996; Sapolsky, 2000).
When measuring brain structure volumes and calculating heritability
estimates, these variables, having a strong genetic component themselves, should
be taken into account.
C. HERITABILITY ESTIMATES TO DETERMINE CONTRIBUTION OF GENETIC
FACTORS TO STRUCTURAL BRAIN TRAITS
(to study genetics of both natural and disease-related variation)
Twin and adoption studies have shown substantial genetic influences are
involved in the risk of developing schizophrenia (Cannon et al., 1998; Cardno
et al., 2002; Kendler and Diehl, 1993; McGuffin et al., 1984). The identification
of predisposing genes has been hampered by difficulties in detecting nonpenetrant carriers and by uncertainties concerning the extent of locus of heterogeneity (McDonald and Murphy, 2003). By studying the inheritance of
endophenotypes, we can increase the power to detect the genes involved in
clarifying the pathways leading from genetic predisposition to clinical disorder
31
HERITABILITY OF STRUCTURAL BRAIN TRAITS
(Gottesman and Gould, 2003). One of the most extensively studied endophenotypes
in schizophrenia research is the study of structural brain abnormalities. These structural changes are well established in schizophrenia. The most robust findings from
meta-analyses of these structural brain abnormalities is increased total ventricular
volume, reduction of whole brain and intracranial volume, reduced hippocampal and
amygdalar volume (Lawrie and Abukmeil, 1998; Ward et al., 1996; Wright et al., 2000).
These brain abnormalities are also present in unaffected family members of patients
with schizophrenia, to a less degree, but significantly more than in controls according
to a meta-analysis reporting on brain changes in unaffected family members of
patients with schizophrenia (Boos et al., 2007; van Haren et al., 2004).
One hypothesis in using these structural abnormalities as endophenotypes, in
order to get closer to the action of the genes, suggests that there might be an
overlap in the neurodevelopmental genes responsible for both the volume change
and the development of the illness, that is, the existence of a genetic correlation.
Schizophrenia may involve genetically determined pathological processes of early
brain development which continue to unfold as the brain matures through
neuronal loss and synaptic pruning during adolescence. Neurodevelopmental
abnormalities then lead to the activation of pathological neural circuits, which
respond to environmental stressors, leading to the emergence of symptoms
(Deutch, 1993). This hypothesis is supported by the finding that MRI abnormalities are present at the onset of the illness, and progress very slowly if at all
(Weinberger, 1995). A contrasting view on the etiology of schizophrenia is that
the relationship between schizophrenia and brain volumes might be environmental
in nature. For example, perinatal trauma has been shown to be an important
determinant of some brain structure abnormalities in schizophrenia (Cannon
et al., 2003; McNeil, 2000; Verdoux et al., 1997).
Defining to what degree the development of these brain structures is of genetic
origin, that is, the heritability of brain structures, and using these brain structures
as an endophenotype may be helpful in better identifying the action of these
neurodevelopmental genes. Heritability estimates for brain structures in healthy
persons and persons with schizophrenia will help to differentiate between disease
related alterations in brain morphology and genetic influences (heritability).
Twin studies are considered important in investigating genetic influences on
variation in human brain morphology in healthy individuals and those with
schizophrenia. The twin model is particularly helpful in determining the relative
contribution of genetic, common, and unique environmental influences on variation in brain structures (Posthuma and Boomsma, 2000). Heritability estimates of
brain structure are usually based on data from monozygotic twins (MZ, who are
nearly genetically identical) and dizygotic twins (DZ, who share on average 50%
of their segregating genes). If heritability estimates are based on the assumption
that for a certain brain measure, MZ twin pairs will resemble each other more
closely than DZ twin pairs, it can be inferred that variation of the brain measure is
32
KAYMAZ AND VAN OS
heritable. However, in addition to genetic influences, common (or shared) environmental influences may play a role in explaining resemblances. The effects of
shared environmental factors may be suggested when correlations in DZ twins are
larger than half of the MZ correlation (Boomsma et al., 2002). The effects of
unique environmental factors are obtained from the extent to which MZ twins do
not resemble each other. However, the twin method has been criticized for its
nongeneralizability due to differences in intrauterine and family environment
differences between the twins and compared with singletons (Price, 1950).
Heritability estimates and the search for genes in diseased and nondiseased
humans are difficult due to the impact of environmental factors. Studies in
nonhuman primates (NHPs) might be helpful in elucidating the impact of environmental factors.
D. NONHUMAN PRIMATE MODEL FOR STUDYING BRAIN
NEUROSTRUCTURAL STUDIES
Obstacles in investigating complex traits in human subjects involve the hard-tocontrol and extensive effects of environment, personal development, and medical
history including administrated medications. To a large extent, these problems can
be overcome by applying an animal model. Particularly useful are NHPs due to the
high conservation of anatomical, neurophysiological and cognitive traits as well
genetic sequence with humans. NHP pedigrees have helped studies of various
complex traits including the neruoanatomical due mostly to the availability of
pedigree facilitating genetic studies, established genetic relationships among pedigree animals, well-documented medical and developmental history and uniform
rearing conditions reducing the influence of environmental factors.
To disentangle the possible mechanisms by which genetic and environmental
factors contribute to the morphological abnormalities in the brain animal models
may be used, specifically NHPs, for example, monkey brains, in which the
influence of environment is reduced. The evolution of cognitive function in
hominoids, for example, depends largely on our understanding of the organization of the frontal lobes in extant humans and apes (Semendeferi et al., 1997).
There are studies with findings showing similarities to human brains, such as the
possibility of microstructural plasticity in the NHP hippocampus (Sapolsky et al.,
1990; Siegel et al., 1993; Uno et al., 1989). Other studies show the impact of
environmental factors on the volumes of different brain structures, such as the
affect of differential rearing on the corpus callosum size in rhesus monkeys
(Sánchez et al., 1998). Studies on brain characteristics and the genetic variation
of regional brain morphologies of NHPs will help us understand the meaning of
the brain characteristics in humans as well. They will also help us determine
whether certain assumptions, such as those about certain brain characteristics
33
HERITABILITY OF STRUCTURAL BRAIN TRAITS
being strictly unique to humans, are correct. For example, it is often claimed that
the frontal lobe is disproportionately larger in humans than in other species, but
conflicting reports exist on this issue. Results of a study by Semendeferi et al. (1997)
indicate that although the absolute volume of the brain and the frontal lobe is the
largest in humans, the relative size of the frontal lobe is similar across hominoids
and humans do not have a larger frontal lobe than expected from a primate brain
of human size. Other important functional parts of the brain, like area 10 in
cortical areas of the frontal lobe, which is involved in higher cognitive functions,
also seem to form the frontal pole in chimpanzee, bonobo, orangutan, macaque,
and gibbon brains. Area 10 has similar cytoarchitectonic features in the hominoid
brain, but aspects of its organization vary slightly across species, including relative
width of its cortical layers and the space available for connections (Semendeferi
et al., 2001). Although some features of the human brain, like the asymmetric
Broca’s area and the planum temporal in Wernicke’s posterior receptive language
area in the left hemisphere, were thought to be unique to humans, studies in great
apes (Cantalupo and Hopkins, 2001) and chimpanzees (Gannon et al., 1998) show
the same anatomic hemispheric asymmetry in the Broca’s area and planum
temporal, respectively.
II. Heritability of Brain Structure Phenotypes
Heritability estimates allow for determining which neuroanatomical measures
are traits with marked genetic components and could therefore be interesting both
from the perspective of natural variation in brain related processes in population
and as an attractive endopehnotype for schizophrenia studies. Investigating heritability in various aspects in healthy individuals, in relation to schizophrenia and in
the NHP model, may answer two essential questions: (1) to what degree are brain
volumes genetically determined? and, (2) which brain volumes are mostly heritable
and therefore useful as endophenotypes for genetic studies?
A. HERITABILITY OF BRAIN STRUCTURES IN HEALTHY PERSONS
There are 24 studies reporting on heritability of brain structures in healthy
persons. These studies are listed in Tables I and II, where the measured brain
structures, the heritability rates of these brain structures or regions (denoted as
h2 or as ICC or as Risch’s ), and the method of calculating heritability
(in superscript) is given, besides the sample size/mean age of the persons (twins,
singletons, family members)/sex ratio/other variables, whenever available and
mentioned in the articles are reported in the table as well, with the study ID and
34
35
Gray matter volume
0.94
0.97
Total brain volume
0.91
0.55–0.85
0.82
0.99
0.82
0.82
0.91
0.90
0.99
0.89
0.91
0.88
0.80
0.938
Cranial, intracranial volume
Brain structure
Heritability of
brain structure
(h2, ICC or )
54 MZ, 58 DZ twin pairs, 34 siblings
12 MZ, 12 controls (all M)
54 MZ, 58 DZ twin pairs, 34 siblings
90 MZ (52 M, 38 F), 37 DZ (22 M, 15 F) twin pairs,
158 unrelated singletons (11.5 years)
107 MZ and DZ twin pairs (9 years)
49 MZ, 65 DZ twin pairs (28 years)
10 MZ (34 years), 9 DZ (23 years) twin pairs
23 MZ (12 M), 23 DZ (16 M) twin pairs, all reading
disability, C (NRD) ¼ 9 MZ (4 M), 9 DZ (4 M) twin pairs
54 MZ, 58 DZ twin pairs, 34 siblings
12 MZ, 12 controls (all M)
90 MZ (52 M, 38 F), 37 DZ (22 M, 15 F) twin pairs,
158 unrelated singletons (11.5 years)
107 MZ and DZ twin pairs (9 years)
74 MZ, 71 DZ twin pairs (all M, 68–79 years)
54 MZ, 58 DZ twin pairs, 34 siblings
44 MZ, 40 DZ twin pairs (all M, 68–78 years)
1881 individuals: 1330 genetically related (60 years)
Sample size and age, gender,
whenever different
TABLE I
STRUCTURAL BRAIN PHENOTYPES IN HEALTHY PERSONS WITH HIGH HERITABILITY
(h2 BETWEEN 0.70 AND 0.99)
Peper et al. (2009)a
Hulshoff Pol et al. (2006)a
Baaré et al. (2001a)a
White et al. (2002)b
Posthuma et al. (2002b)b
Wallace et al. (2006)a
Peper et al. (2009)a
Baaré et al. (2001a)a
White et al. (2002)b
Wallace et al. (2006)a
Bartley et al. (1997)a
Pennington et al. (2000)b
Carmelli et al. (1998)a
Baaré et al. (2001a)a
Sullivan et al. (2001)a
Atwood et al. (2004)a
Study ID
36
0.73, 0.90
0.89,
0.906, 0.705
p ¼ 0.02
0.79
0.94
0.90
0.85
0.93
0.83
0.84, 0.75, 0.75
0.80
0.90–0.95
0.84, 0.88, 0.86
CC
Midline structures
Cau, put, thal
Cau
Middle frontal volume
FL, TL, PL
0.91
0.69–0.82
0.71
0.88
0.98
0.97
0.85
Size, shape
Length, area,
AC-PC distance
Area
CC
CC
CC
CC
Calosal vol.
White matter
volume/hyper-intensity
10 MZ, 10 DZ twin pairs (50–70 years)
90 MZ, 37 DZ twin pairs, 158 unrelated
singletons ( 11.5 years)
107 MZ and DZ twin pairs (9 years)
12 MZ, 12 controls (all M)
90 MZ, 37 DZ twin pairs, 158 unrelated
singletons ( 11.5 years)
Tramo et al. (1998)b
Sullivan et al. (2001)a
Scamvougeras et al. (2003)b
Pfefferbaum et al. (2004)a
Wallace et al. (2006)a
10 MZ twin pairs
44 MZ, 40 DZ twin pairs (all M, 68–78 years)
14 MZ, 12 DZ twin pairs (27 years)
34 MZ, 37 DZ (all M, 76 years)
90 MZ, 37 DZ twin pairs, 158 unrelated
singletons ( 11.5 years)
107 MZ and DZ twin pairs (9 years)
(continued )
Thompson et al. (2001)b
Wallace et al. (2006)a
Peper et al. (2009)a
White et al. (2002)b
Wallace et al. (2006)a
Peper et al. (2009)a
Oppenheim et al. (1989)b
Biondi et al. (1998)a
Peper et al. (2009)a
Hulshoff Pol et al. (2006)a
Carmelli et al. (1998)a
Baaré et al. (2001a)a
White et al. (2002)b
Posthuma et al. (2002b)b
Wallace et al. (2006)a
5 MZ twin pairs, 10 unrelated controls
7 MZ twin pairs
74 MZ, 71 DZ twin pairs (all M, 68–79 years)
54 MZ, 58 DZ twin pairs, 34 siblings
12 MZ, 12 controls (all M)
54 MZ, 58 DZ twin pairs, 34 siblings
90 MZ (52 M, 38 F), 37 DZ (22 M, 15 F) twin pairs,
158 unrelated singletons ( 11.5 years)
107 MZ and DZ twin pairs (9 years)
49 MZ, 65 DZ twin pairs ( 28 years)
37
0.83
0.83
Lateral ventricle
Cerebral ventricle
Posthuma et al. (2000)a
White et al. (2002)b
Pennington et al. (2000)b
Pfefferbaum et al. (2004)a
Reveley et al. (1984)a,b,c
34 MZ, 37 DZ (all M, 76 years)
18 MZ, 21 DZ twin pairs (37 years)
Study ID
43 MZ, 37 DZ twin pairs, 34 siblings
12 MZ, 12 controls (all M)
23 MZ (12 M), 23 DZ (16 M) twin pairs, all reading
disability, C (NRD) ¼ 9 MZ (4 M), 9 DZ (4 M)
twin pairs
Sample size and age, gender,
whenever different
Summary of published studies by August 2009 (PubMed and Medline citations) on the heritability of brain structural phenotypes including study results from
healthy monozygotic (MZ) and dizygotic (DZ) twins, singletons (sibs), and in some studies healthy controls (C). Heritability is denoted as h2 or as intraclass
correlation factor (ICC) or as Risch’s . A higher ICC for MZ twins then for DZ twins indicates a genetic component to these structures, but large confidence
intervals indicate that the similarities measured are a rough estimates of the degree to which a brain structure is genetically controlled. Another way of denoting
the heritability is by s ¼ Risch’s ¼ a measure of heritability. For Risch’s , s ¼ 1 is the expected value for nonfamilial traits, whereas s > 2.0 is considered
evidence for familiality. h2 or ICC or Risch’s ¼ heritability rate.
TB, total brain volume; WM, total white matter volume; GM, total gray matter volume; WM, white matter volume; GM, gray matter volume; CC, corpus
callosum; AC-PC distance, the distance between anterior and posterior commissure; CB, cerebellar volume; FL, frontal lobe; TL, temporal lobe; PL, parietal
lobe; OL, occipital lobe; OC, occipital cortex; TH, temporal horn volume; LV, lateral ventricle volume; TV, third ventricle; HC, hippocampal volume; Parahip,
parahippocampal gyrus; Cau, caudate nucleus; Thal, thalamus; SS or SL, sulcus shape or sulcus length; put, putamen; Hpt, hypothalamus; PHC gyrus,
parahippocampal gyrus; MZ, monozygote; DZ, dizygote; M, male, F, female; Sibs, siblings; n.a., not available; n.s., not significant.
a
SEM, structural equitation modeling., bFalconer method., cAnalysis of variance estimate of heritability.
Subcortex
0.88
0.99
0.70
Cerebellum
Brain structure
Heritability of
brain structure
(h2, ICC or )
TABLE I (continued )
38
0.66, 0.56
0.59, 0.41 (n.s.)
0.41 (n.s.), 0.47
0.28 (n.s.), 0.69
0.66
0.61
0.40
0.56
0.30–0.57
0.553
0.612
MZ cc:0.21, MZ disc: 0.41
0.62
0.60
0.620
0.49
Total brain volume
Hippocampus
Neocortex
Cerebral Cortex
WM hyperintensity
Planum Temporale asymmetry
Brain Parenchyma
Temporal Horn
CC perimeter
Cerebellum
Heritability of brain
structure (h2, ICC or )
Frontal lobe (L, R), Temporal lobe
(L, R), Parietal lobe (L, R)
Occipital lobe (L, R)
Brain structure
Carmelli et al. (1998)b
Sullivan et al. (2001)b
Biondi et al. (1998)d
Wallace et al. (2006)b
74 MZ, 71 DZ twin pairs (all M, 68–79 years)
44 MZ, 40 DZ twin pairs (all M, 68–78 years)
7 MZ twin pairs
90 MZ, 37 DZ twin pairs, 158 unrelated singletons
( 11.5 years)
(continued )
Atwood et al. (2004)b
1) Eckert et al. (2002)a
2) Steinmetz et al. (1995)a
Schmitt et al. (2008)b
Pennington et al. (2000)a
Sullivan et al. (2001)b
Wright et al. (2002)b
Geschwind et al. (2002)a,b,c
Geschwind et al. (2002)a,b,c
Study ID
1881 individuals: 1330 genetically related ( 60 years)
27 MZ, 12 DZ win pairs ( 27 years)
10 MZ concordant and 10 MZ discordant twin pairs for
handedness (26 and 22 years)
23 MZ (12 M), 23 DZ (16 M) twin pairs, all reading
disability, C (NRD) ¼ 9 MZ (4 M), 9 DZ (4 M)
twin pairs
107 MZ, 47 DZ twin pairs, siblings of twins 64,
singletons 228
44 MZ, 40 DZ twin pairs (all M, 68–78 years)
9 MZ, 10 DZ twin pairs ( 23 years)
72 MZ, 67 DZ twin pairs (71 years)
72 MZ, 67 DZ twin pairs (71 years)
Sample size
TABLE II
STRUCTURAL BRAIN PHENOTYPES IN HEALTHY PERSONS WITH MODERATE HERITABILITY (h2 BETWEEN 0.40 AND 0.70), LOW HERITABILITY
(h2 BETWEEN 0.00 AND 0.40) AND NO SIGNIFICANT HERITABILITY (h2 ¼ 0 AND DENOTED AS N.S.)
39
0.377 (n.s.)
0.31
Heritability of brain
structure (h2, ICC or )
27 MZ, 12 DZ twin pairs (6–16 years)
90 MZ, 37 DZ twin pairs, 158 unrelated singletons
( 11.5 years)
Sample size
Eckert et al. (2002)a
Wallace et al. (2006)b
Study ID
Summary of published studies by August 2009 (PubMed and Medline citations) on the heritability of brain structural phenotypes including study results from
healthy monozygotic (MZ) and dizygotic (DZ) twins, singletons (sibs), and in some studies healthy controls (C). Heritability is denoted as h2 or as intraclass
correlation factor (ICC) or Risch’s . A higher ICC for MZ twins then for DZ twins indicates a genetic component to these structures, but large confidence
intervals indicate that the similarities measured are a rough estimates of the degree to which a brain structure is genetically controlled. Another way of denoting
the heritability is by s ¼ Risch’s ¼ a measure of heritability. For Risch’s , s ¼ 1 is the expected value for nonfamilial traits, whereas s > 2.0 is considered
evidence for familiality. h2 or ICC or Risch’s ¼ heritability rate.
TB, total brain volume; WM, total white matter volume; GM, total gray matter volume; WM, white matter volume; GM, gray matter volume; CC, corpus
callosum; AC-PC distance, the distance between anterior and posterior commissure; CB, cerebellar volume; FL, frontal lobe; TL, temporal lobe; PL, parietal
lobe; OL, occipital lobe; OC, occipital cortex; TH, temporal horn volume; LV, lateral ventricle volume; TV, third ventricle; HC, hippocampal volume; Parahip,
parahippocampal gyrus; Cau, caudate nucleus; Thal, thalamus; SS or SL, sulcus shape or sulcus length; put, putamen; Hpt, hypothalamus; PHC gyrus,
parahippocampal gyrus; MZ, monozygote; DZ, dizygote; M, male, F, female; Sibs, siblings; n.a., not available; n.s., not significant.
a
Falconer method., bSEM, structural equitation modeling., cAnalysis of variance estimate of heritability., dVisual comparison of the 3D surface rendering
models of the brains and manual measurements of the AC-PC distance by six blinded observers.
Other (n.s. or low h2):
SS or SL
LV
Brain structure
TABLE II (continued )
HERITABILITY OF STRUCTURAL BRAIN TRAITS
year of publication. Magnetic resonance imaging (MRI scans) brain scans were
obtained from MZ and dizigotic (DZ) healthy twin pairs. The inclusion of
monozygotic, dizygotic, singletons, and family members, but also unrelated controls is done to assess the extent of genetic control over these structures. Within
pair similarities for certain brain structures can be compared for each group, for
example within pair similarity for brain structures with high heritability rates such
as the total brain volume increases as pair members are more closely related
genetically (monozygotic > siblings (included dizygotic) > unrelated controls).
In one study healthy controls were also included (Oppenheim et al., 1989). In five
studies, unrelated and related singletons were included as well (Baaré et al., 2001a;
Posthuma et al., 2000, 2002b; Schmitt et al., 2008; Wallace et al., 2006), and in one
study (Atwood et al., 2004) family members were also included. In three studies the
heritability rates were investigated only in a male population (Carmelli et al., 1998;
Eckert et al., 2002; Sullivan et al., 2001), and the remaining studies consisted of a
mixed population. The age range for most of the studies was between 23 and 80
years, except for one study (Peper et al., 2009) in which 9-year-old children were
included.
There were 37 structures measured in total. Some structures were measured in
multiple studies, such as the total brain volume (six studies), total cranial/intracranial volume (four studies), white matter (seven studies), white matter hyperintensity
(two studies), gray matter (six studies), corpus callosum (eight studies), cerebellar
volume (three studies), frontal/temporal/parietal lobe (two studies), lateral ventricle
(two studies), amygdalar and caudate (two studies), sulcus shape/length (two studies), planum temporal asymmetry (two studies), and ventricular volume (three
studies). The rest of the structures were measured in a single study. In one study
(Hulshoff Pol et al., 2006) gray matter and white matter volume was measured for
different brain parts such as superior frontal, medial frontal, postcentral gyrus,
posterior cingulated, and Heschl’s gyrus for the left and the right part. In Table I,
the heritability rates ranging from 0.55 to 0.85 for gray matter and 0.69–0.82 for
white matter volume are given. In another study (Schmitt et al., 2008) various parts
of the cerebral cortex were investigated and in Table II heritability rates between
0.30 and 0.57 for the cerebral cortex are recorded.
1. Results on the Heritability of the Reviewed Studies
Gross brain structures show higher heritability rates than specific structures.
High heritability rates ranging from 0.70 to 0.99 are characteristic to intracranial
volume, whole brain volume, total gray matter volume, total white matter volume, calosal volume, middle frontal volume, midline structures, and the volumes
of both the right and left hemispheres (Table I). Studies including DZ, resulting in
a full twin study (Bartley et al., 1997; Carmelli et al., 1998; Pennington et al., 2000),
reported heritability of >0.90 for total brain volume. A moderate heritability rate
was found on the four lobes investigated, such as frontal, temporal, occipital, and
40
KAYMAZ AND VAN OS
parietal lobe, ranging from 0.41 to 0.70 (Geschwind et al., 2002), but conflicting
results were reported on the lobes by Wallace et al. (2006) for the frontal, temporal
and parietal lobe, resulting in high h2 scores between 0.84 and 0.88. Moderate
heritability rates are also reported for temporal horn volume, brain parenchyma,
hippocampus, neocortex, white matter hyperintensity, and planum temporal
asymmetry ranging from 0.60 to 0.63. Studies showing a moderate or a low
heritability of brain structures are listed in Table II.
2. Effects of Gender, Age, Handedness, and Development in Time
Some studies have investigated variables that could have an influence on
heritability rates. In a study by Steinmetz et al. (1995), the effect of gender on
measuring the midsagittal area of the corpus callosum and total brain volume in
120 healthy young adults was shown, finding that the forebrain volume-adjusted
size of the corpus callosum was larger in women than in men and that handedness
had no effect on this measurement. However, an effect of handedness on cerebral
asymmetry was found in two studies (Eckert et al., 2002; Steinmetz et al., 1995)
reporting a lack of cerebral asymmetry in persons who were left handed. This
might be explained by the hypothesis that early ‘‘epigenetic’’ events that take
place during embryogenesis may contribute to a significant variability in the
development of the anatomofunctional laterality of the cerebral hemispheres.
In a study by Peper et al. (2009), children (n ¼ 47) with signs of secondary sexual
characteristics (based on Tanner stages) were compared to children without these
signs and the heritability of brain structures was estimated using MRI scans. They
found high heritability for most gross brain structures (see Table I) and regional
brain structures. They also found that the onset of secondary sexual characteristics of puberty were associated with decreased frontal and parietal gray matter
densities that accompany the transition of our brains from childhood into adulthood, still showing high heritability for brain structures such as global brain
volumes, white matter density in fronto-occipital and superior longitudinal fascicles and gray matter density of (pre)frontal and temporal areas. Only one study
(Pfefferbaum et al., 2004) investigated the heritability rates examined over time
and found no evidence for new genetic variance on the h2 scores of corpus
callosum and lateral ventricles recalculated after 4 years. The heritability of
surface morphology, which is formed later in development and assumed to be
under more environmental influence than the deeper structures (formed earlier in
life and under more genetic control) indicate that sulcal depth effect was stronger
for related twins. These results show that the shape of the deeper sulci is more
strongly predetermined than that of superficial sulci, for example, sulcal depth
having an influence on sulcal similarity. Also, subjects with similar brain shape
have more similar patterns of their sulci, suggesting both features are the result of
ontogenetically related processes (Bartley et al., 1997; Biondi et al., 1998;
41
HERITABILITY OF STRUCTURAL BRAIN TRAITS
Lohmann et al., 1999; White et al., 2002). This anatomical similarity was also
found for the corpus callosum shape and size (Oppenheim et al., 1989).
3. Conflicting Results on Some Brain Structures
Two studies (Posthuma et al., 2000; White et al., 2002) report high h2 rates for
cerebellar volume and one study (Wallace et al., 2006) reports a moderate h2 rate.
This might be due to the age difference of the participants, and thus, the ongoing
development of the brain. In the first two studies, adults were included, whereas the
latter study, included adolescents (mean age of 11.5 years). A low heritability rate for
lateral ventricle volume was found in one study (Wallace et al., 2006), but a high h2
score was reported in another (Pfefferbaum et al., 2004), and an h2 of 0.00 is
recorded by Wright et al. (2002). The differences in these studies might be due to
the differences in age of the participants, but could also be due to the differences in
sample size. The mean age of the participants in the study by Wallace et al. (2006)
was 11.5 years, reflecting results of brains in development. Additionally, the sample
size in the study by Wright et al. (2002) is small compared to the bigger sample sizes
of the other two studies. Conflicting results also exist on the heritability of sulcus
shape and length, with one study indicating moderate heritability rate (Bartley et al.,
1997) and not reaching significance in another (Eckert et al., 2002). There is a
difference in the mean age between the studies, resulting in an older mean age
(23 years) in the study by Bartley et al. (1997) and younger (6–16 years) in the study
by Eckert et al. (2002). However, it is unclear what might be of influence in the
differences between these studies, because different statistical heritability estimates
were used in both. Other conflicting results are found for the ventricle volume,
having a high h2 rate in one study (Reveley et al., 1984) and not being significant in
other studies (Baaré et al., 2001a; White et al., 2002). In the study by Reveley et al.
(1984), three statistical ways of calculating heritability rates were used, whereas the
other two studies used one of the methods (both different from each other).
The differences between the studies with conflicting results might be a result of
the differences in sample size, age, gender of the population (only male participants vs. mixed group), statistical method applied to calculate the heritability rates
and the way measurements of the brain structures is done. As for the latter, some
studies used the manual measurement of brain structures (region or VOI) and
others automated (VBM, deformation based morphometry).
Seven studies used the Falconer method of heritability calculation and
15 studies used the structural equitation modeling (SEM), in which the unique
environmental factors are also included. Two studies (Geschwind et al., 2002;
Reveley et al., 1984) used all three methods of heritability calculation, reaching
significance in both studies for the three methods.
Heritability estimates show considerable differences between gross and specific brain structures. High heritability is characteristic for gross brain structures,
such as intracranial volume, whole brain volume, cerebellar volume, total gray
42
KAYMAZ AND VAN OS
matter and white matter volume, and calosal volume, with effect sizes in the high
range. Moderate effect sizes have been found for structures such as the heritability
of the different lobes, sulcus shape or length, white matter hyperintensity, hippocampal volume, and other brain structures like brain parenchyma, middle frontal
regions, temporal horn and planum temporal asymmetry. The highly heritable
gross brain structures are also the brain structures that are examined in multiple
studies. The volume measurements of these brain structures and thus, the heritability rates, are easier to calculate than the specific, middle brain, deeper or small
brain structures in both the manual and automated quantitative measurement
techniques. From the reviewed literature, these brain structures with high and
moderate heritability rates seem to be useful as endophenotypes.
B. HERITABILITY OF BRAIN STRUCTURES IN PATIENTS WITH SCHIZOPHRENIA
COMPARED TO HEALTHY CONTROLS
Heritability calculations of the brain structures of patients with schizophrenia
compared to healthy controls is very complicated, due to the impact of disease
related factors on brain changes, together with the effect of medication intake and
the use of drugs (Kanaan et al., 2009; Rais et al., 2008; van Haren et al., 2007). Some
studies show an increase in brain structure volumes and others a decrease. However,
use of medication does not account for the total changes in the brain structures of
patients with schizophrenia, since there are also studies showing brain changes in
medication naı̈ve patients (Cahn et al., 2001; Okugawa et al., 2007; Weinberger,
1995). Using twins concordant and discordant for schizophrenia in these studies
would help elaborate the disease’s effect on the brain changes to a certain degree.
Until now, there have been nine studies published on the heritability of brain
structures in patients with schizophrenia compared to healthy controls. These
studies are listed in Table III. The studies report on the heritability rates of brain
structures in twins concordant and discordant for schizophrenia, monozygotic, and
DZ, compared to healthy twins. These studies included adult patients, females, and
males. Some brain structures have been studied in multiple studies, such as the total
brain volume (four studies), gray matter volume (two studies), hippocampal volume
(four studies), and the rest of the brain structures listed on the table in single studies.
High heritability for the intracranial volume, total gray and white matter
volume, ranging from 0.72 to 0.89 was found (Table III). The overall gray matter
volume was also found to be highly heritable in monozygote twins, both in the
control and in the group of patients with schizophrenia. The heritability of total
brain volume was found to be high in two studies (Hulshoff Pol et al., 2004; Rijsdijk
et al., 2005) and high in patients with schizophrenia, but moderate in healthy
controls in another study (Narr et al., 2002). In one study (Rijsdijk et al., 2005), a
low heritability rate was found for the hippocampus, third and lateral ventricle.
43
44
(1) SCZ: MZ: 0.86, DZ: 0.06
C: MZ: 0.93, DZ: 0.36
(2) 0.86
TB volume
(1) SCZ: MZ: 0.73, DZ:0.05
C: MZ: 0.83, DZ ¼ 0.32
(2) 0.72
0.72
(1) FL; SCZ: MZ:0.84, DZ: 0.26
C: MZ: 0.93, DZ: 0.00
(2) FL: 0.76, TL: 0.79
GM volume
WM volume
Frontal lobe
Parietal lobe
(4) 0.66
(3) 0.88
(1) SCZ: MZ:0.93, DZ:0.45
C: MZ:0.86, DZ: 0.35
(2) 0.89
Heritability of brain
structure (h2, ICC or )
Intracranial or cranial volume
Brain structure
(2) 9 MZ, 9 DZ disc for SCZ, 14 MZ and
13 DZ healthy twin pairs
(1) 15 MZ, 15 same sex DZ disc for SCZ,
29 healthy twin pairs
11 MZ, 11 DZ disc for SCZ, 11 MZ and
11 DZ C twin pairs
(1) 20 MZ, 20 DZ twin pairs with SCZ, 20 MZ
and 20 DZ healthy twin pairs ( 48 years)
(2) 11 MZ, 11 DZ disc for SCZ,
11 MZ and 11 DZ C twin pairs
(1) 15 MZ, 15 same sex DZ disc for SCZ,
29 healthy twin pairs
(2) 11 MZ, 11 DZ disc for SCZ, 11 MZ and
11 DZ C twin pairs
(3) 14 MZ cc for SCZ and 10 MZ disc for SCZ,
17 MZ C pairs, 22 disc sibling pairs,
3 cc sibling pairs, 114 healthy C
(4) 9 MZ, 9 DZ disc for SCZ, 14 MZ and
13 DZ healthy twin pairs
(1) 15 MZ, 15 same sex DZ disc for SCZ,
29 healthy twin pairs
(2) 11 MZ, 11 DZ disc for SCZ, 11 MZ
and 11 DZ C twin pairs
Sample size
Study ID
(continued )
(2) Brans et al. (2008)a
(1) Baare et al. (2001b)a
Hulshoff Pol et al. (2004)a
(2) Hulshoff Pol et al. (2004)a
(1) Cannon et al. (2002)b
(4) Brans et al. (2008)a
(3) Rijsdijk et al. (2005)a
2) Hulshoff Pol et al. (2004)a
(1) Baare et al. (2001b)a
(2) Hulshoff Pol et al. (2004)a
(1) Baare et al. (2001b)a
TABLE III
HERITABILITY OF STRUCTURAL BRAIN PHENOTYPES IN PERSONS WITH SCHIZOPHRENIA COMPARED TO HEALTHY CONTROLS
45
(1) SCZ: MZ:0.36, DZ: 0.17
C: MZ: 0.71, DZ: 0.27
(2) 0.33
(1) SCZ: MZ: 0.67, DZ:0.18
C: MZ: 0.73, DZ: 0.00
(2) 0.05
Third ventricle
Lateral ventricle
(3) L: 0, R: 0 (s)
(1) SCZ: MZ:0.91, DZ:0.33
C: MZ: 0.54, DZ: 0.74
(2) L: 4.93, R: 8.50 (s)
(3) 0.36
(4) L:3.10, R: 2.33 (s)
(1) SCZ: MZ: 0.68, DZ: 0.00
C: MZ: 0.73, DZ: 0.00
(2) SCZ: 0.42, C: 0.71
Heritability of brain
structure (h2, ICC or )
Corpus callosum
Hippocampus
Brain structure
Sample size
(1) 15 MZ, 15 same sex DZ disc for SCZ,
29 healthy twin pairs
(2) 14 MZ cc for SCZ and 10 MZ disc
for SCZ, 17 MZ C pairs, 22 disc sibling
pairs, 3 cc sibling pairs, 114 healthy C
(3) SCZ: 169 patients, unaffected siblings:
183, controls: 221
(1) 15 MZ, 15 same sex DZ disc for SCZ,
29 healthy twin pairs
(2) 14 MZ cc for SCZ and 10 MZ disc for
SCZ, 17 MZ C pairs, 22 disc sibling
pairs, 3 cc sibling pairs, 114 healthy C
(1) 10 MZ, 10 DZ twin pairs disc for SCZ
and 10 MZ, 10 DZ unaffacted co twin pairs
(2) SCZ: 169 patients, unaffected siblings:
183, controls: 221
(1) 15 MZ, 15 same sex DZ disc for SCZ,
29 healthy twin pairs
(2) 7 MZ cc and 16 MZ disc and 28 DZ
disc twin pairs for SCZ, 28 MZ and
26 DZ healthy control twin pairs
(3) 14 MZ cc and 10 MZ disc for SCZ
(4) SCZ: 169 patients, unaffected siblings:
183, controls: 221
TABLE III (continued )
(3) Goldman et al. (2007)
(2) Rijsdijk et al. (2005)a
(1) Baare et al. (2001b)a
(2) Rijsdijk et al. (2005)a
(1) Baare et al. (2001b)a
(2) Goldman et al. (2007)
(1) Narr et al. (2002)b
(3) Rijsdijk et al. (2005)a
(4) Goldman et al. (2007)
(2) van Erp et al. (2004)a
(1) Baare et al. (2001b)a
Study ID
46
SCZ: 169 patients, unaffected siblings:
183, controls: 221
L: 0, R: 0.87 (s)
Dorsal striatum
Goldman et al. (2007)
Baare et al. (2001b)a
Koolschijn et al. (2007)a
Summary of published studies by August 2009 (PubMed and Medline citations) on the heritability of brain structural phenotypes with study results from twins
concordant (cc) and discordant (disc) for schizophrenia (SCZ), singletons, and healthy controls (C). Heritability is denoted as h2 or as intraclass correlation factor
(ICC). A higher ICC for monozygotic (MZ) twins then for dizygotic (DZ) twins indicates a genetic component to these structures, but large confidence intervals
indicate that the similarities measured are a rough estimates of the degree to which a brain structure is genetically controlled.
Another way of denoting the heritability is by s ¼ Risch’s ¼ a measure of heritability. For Risch’s , s ¼ 1 is the expected value for nonfamilial traits,
whereas s > 2.0 is considered evidence for familiality. h2 or ICC or Risch’s ¼ heritability rate.
TB, total brain volume; WM, total white matter volume; GM, total gray matter volume; WM, white matter volume; GM, gray matter volume; CC, corpus
callosum; AC-PC distance, the distance between anterior and posterior commissure; CB, cerebellar volume; FL, frontal lobe; TL, temporal lobe; PL, parietal
lobe; OL, occipital lobe; OC, occipital cortex; TH, temporal horn volume; LV, lateral ventricle volume; TV, third ventricle; HC, hippocampal volume; Parahip,
parahippocampal gyrus; Cau, caudate nucleus; Thal, thalamus; SS or SL, sulcus shape or sulcus length; put, putamen; Hpt, hypothalamus; PHC gyrus,
parahippocampal gyrus; MZ, monozygote; DZ, dizygote; M, male; F, female; Sibs, siblings; n.a., not available; n.s., not significant.
a
SEM, structural equitation modeling.
b
Falconer method.
15 MZ, 15 same sex DZ disc for SCZ,
29 healthy twin pairs
SCZ: MZ: 0.34m DZ: 0.37
C: MZ: 0.70, DZ: 0.10
PHC gyrus
11 MZ, 11 DZ same sex twin pairs, disc for
SCZ and 11 MZ, 11 DZ same sex
healthy twin pairs
MZ: ICC ¼ 0.58, p < 0.05, DZ: 0.07.
Correction for TB!n.s.
Hypothalamus
KAYMAZ AND VAN OS
Low heritability rates were also found in the group of DZ twins discordant for
schizophrenia (Narr et al., 2002), for the overall gray matter volume in control DZ
twins and in DZ twins with schizophrenia. There was no significant heritability
found for the parahippocampal gyrus, third and lateral ventricle (Baaré et al.,
2001b) or for the hypothalamus (Koolschijn et al., 2007). The study by Brans et al.
(2008) reports on whether genetic (heritability) and/or environmental factors are
associated with progressive brain volume changes in schizophrenia. They found
that significant decreases over time in whole brain, frontal, and temporal lobe
volume in patients with schizophrenia and in their unaffected co-twins compared
with control twins revealed significant additive influences on the correlations
between schizophrenia liability and progressive whole brain (0.66), temporal
(0.79), and frontal (0.76) lobe changes.
The samples in these studies consist of both healthy MZ and DZ twins and
twins concordant and discordant for schizophrenia, with the exception of three
studies (Baaré et al., 2001b; Hulshoff Pol et al., 2004; Koolschijn et al., 2007) where
twins concordant for schizophrenia were not included, resulting in a sample
group where environmental factors played a more important role in the development of the disorder and a genotypic estimation of the true population genetic
variance was not possible.
Three studies (Cannon et al., 2002; Narr et al., 2002) used the Falconer method
of heritability estimates and four studies used the SEM method (Brans et al., 2008;
Hulshoff Pol et al., 2004; Rijsdijk et al., 2005; Van Erp et al., 2004). The study by
Koolschijn et al. (2007) used the intraclass correlation factor (ICC) to estimate the
heritability rates of the hypothalamus. One study, by Goldman et al. (2007), used
ICC and also Risch’s relative risk for siblings (s) for heritability measurements of
each region. Risch’s relative risk for siblings (s), was calculated to quantitatively
measure the likelihood that siblings of patients with volumetric reductions or
increases would express the phenotype themselves. Because this parameter
depends on the presence or absence of the phenotype, cutoffs were defined on
the basis of the distribution of the volumes in patients.
One complicating factor in the heritability estimates of brain structures in
twin studies is that the twin discordant for schizophrenia also shows, to a milder
degree, brain changes seen in their co-twins with schizophrenia. The unaffected
twins show lower brain volume, lower hemispheric volumes and/or frontal lobe
gray matter volume decrease. These observations may suggest the involvement of
genetic factors in cortical gray matter deficits in patients with schizophrenia
(Baaré et al., 2001b; Reveley et al., 1984; Suddath et al., 1990). As suggested by
Cannon et al. (2002), it is important that these studies include an evaluation of the
regional cortical variation and that individual differences in gyral/sulcal patterning
and shape are taken into account. It is not clear to what extent these changes are
similar to the changes in their co-twins who have schizophrenia. A distinction
between these possibly genetic differences and the disease related changes in the
47
HERITABILITY OF STRUCTURAL BRAIN TRAITS
brain should be done by isolating the topography of nongenetic, disease related
differences in cortical gray matter volume, as suggested by Cannon et al. (2002).
C. HERITABILITY OF BRAIN STRUCTURES IN NONHUMAN PRIMATES (MONKEYS)
Studies in pedigrees, where there is an extensive control of the environmental
factors, are particularly useful in heritability calculations. Studies in NHPs are
therefore a good model for genetic control and could be useful in heritability
calculations as mentioned previously.
There are five studies published on the heritability of brain structures in NHPs
(e.g., monkeys). These studies are listed in Table IV. Results of these studies
strongly support findings from human studies showing the heritability of gross
brain structures in the high effect size range, such as whole brain volume, cerebral
volume, and cranial size. The moderate effect size range for heritability was
reported for cranial capacity, sulcus length, brain weight, hippocampus and
surface area, and gray matter. The number of studies reporting on heritability
of brain structures in NHPs is limited and the sample sizes of monkeys included in
these studies is small. There are only two studies (Mahaney et al., 1993; Rogers
et al., 2007) in which the same monkey type was included, but these studies did not
look at the same brain structure, which makes comparison difficult. Only the
study by Fears et al. (2009) used a big sample size of vervet monkeys and calculated
heritability estimates using manual tracings as well as automated tracings of brain
structures. In this study, high heritability rates are found for gross brain structures,
such as total brain volume, cerebral volume and corpus callosum. In contrast to
other NHP heritability estimations of the hippocampus, this study showed a high
heritability rate for the hippocampus. The findings from the NHP studies support
the findings in human twin studies.
Table IV shows that samples consist of different types of monkeys, making it
difficult to compare the conflicting results between studies on the same brain
structure, such as the hippocampus, where Lyons et al. (2001) find moderate
heritability rates and Fears et al. (2009) find a high heritability rate. There is also
a big difference in the sample sizes.
One other complicating factor is the method of measuring the volume of brain
structures (manual vs. automated or both), particularly, the calculation of heritability of brain structures across the different studies. The method used by Lyons
et al. (2001) consists of obtaining MRI scans of Guyanese monkeys and quantitative estimates of the heritability rates were generated by using the one-way
ANOVA as used to evaluate paternal half siblings effect. Rogers et al. (2007)
used the VBM approach to look at the regional differences in the genetic effects
on brain structure and calculated the heritability estimates. The study by
Cheverud et al. (1990) shows that brain size was defined as endocranial capacity
48
49
(1) 0.82 (p ¼ 0.00022)
(2) 0.99 (SE ¼ 0.06)
0.67, p ¼ 0.01
(1) 0.86, p ¼ 0.0006
(2) 0.98 (SE ¼ 0.09)
0.673, p ¼ 0.0069
0.87 (SE ¼ 0.07)
(1) 0.54 (F ¼ 2.58, p ¼ 0.02)
(2) 0.95 (SE ¼ 0.07)
0.31, p < 0.05
Total brain volume
(1) Baboons: 601; 307 assigned to 25 pedigrees (2–49 members),
294 individual animals
(2) Rhesus Macaques; 179 skeletons, measuring
endocranial capacity and using mother offspring regression
and SDS method
(1) Baboons (papio hamadryas): 109 MRI Scans (VBM approach)
(2) Vervet research colony: 357 (>2 years of age)
Baboons (papio hamadryas): 109 MRI Scans (VBM approach)
(1) Baboons (papio hamadryas): 109 MRI Scans (VBM approach)
(2) Vervet research colony: 357 (>2 years of age)
Baboons (papio hamadryas): 109 MRI Scans (VBM approach)
Vervet research colony: 357 (>2 years of age)
(1) Guyanese (saimiri sciures): 39 MRI scans (1 way ANOVA)
(2) Vervet research colony: 357 (>2 years of age)
Rhesus Macaques; 179 skeletons, measuring
endocranial capacity and using mother offspring
regression and SDS method
Sample size and monkey type
(1) Rogers et al. (2007)
(2) Fears et al. (2009)
Rogers et al. (2007)
(1) Rogers et al. (2007)
(2) Fears et al. (2009)
Rogers et al. (2007)
Fears et al. (2009)
(1) Lyons et al. (2001)
(2) Fears et al. (2009)
Cheverud et al. (1990)
(2) Cheverud et al. (1990)
(1) Mahaney et al. (1993)
Study ID
Summary of published studies by August 2009 (PubMed and Medline citations) on the heritability results of brain structural phenotypes in nonhuman
primates. Heritability is denoted as h2.
SDS, symmetric differences squared method, SE, standard error, MRI, magnetic resonance imaging; VBM, voxel based morphometry method.
Sulcus length
Surface area of the cerebrum
Corpus callosum
Hippocampus
Gray matter volume
Cerebral volume
(2) 0.75, p < 0.005
0.60, p ¼ 0.02
(1) 0.409, p < 0.005
Heritability of brain
structure (h2, ICC or )
(2) Size Cranial capacity
Brain
(1) Weight
Brain structure
TABLE IV
HERITABILITY OF STRUCTURAL BRAIN PHENOTYPES IN NONHUMAN PRIMATES
HERITABILITY OF STRUCTURAL BRAIN TRAITS
and was measured by filling the cranium with mustard seed, then pouring the
enclosed seed into a graduated cylinder. The cube root of cranial capacity was used
in statistical analysis to better fit the data with statistical models. The heritability was
estimated by standard mother–offspring regression (Falconer and Mackay, 1996;
Falconer et al., 1981) and a slight modification of the symmetric-differences-squared
(SDS) method (Bruckner and Slanger, 1986). In the study by Mahaney et al. (1993),
the heritability estimates, specifically of the brain weight, were measured by quantitative genetic analyses on available loge-transformed organ weights. Fears et al.
(2009) used both manual and deformation based volume measurements (automated) to measure volumes of brain structures.
Despite the difference in heritability rates of brain structures between the few
studies published on this topic, the results are nonetheless comparable to the
human studies. The newest publications on this subject in NHP (Fears et al., 2009;
Rogers et al., 2007) have used big sample sizes and also comparable volume
measurements on the MRI scans as seen in human studies. More research on
heritability in NHPs will help us get more insight into the genetic control on brain
structures and will create the potential for future genetic mapping of QTLs
underlying the variance in the heritability of these brain structures.
III. Genes for Brain Structures in Healthy and Persons with Schizophrenia
The dramatically increased brain volume plays an important role in the origin
of our species. In humans, brain volume is a quantitative trait with high heritability (Posthuma et al., 2002b; Thompson et al., 2001). Disturbances in brain development is one of the explanations for neuropsychiatric disorders such as
schizophrenia (Weinberger, 1995), suggesting that developmental genes involved
in the development of the brain may overlap with the causal genes, for example,
potential candidate genes, for schizophrenia (Murray, 1994). However, no susceptibility genes with high effect sizes have been found for schizophrenia. Therefore, most studies have focused on the testing of specific genetic markers in a
known candidate gene for association with the different endophenotypes. In this
study, research using brain structures as neuroanatomical endophenotypes in
healthy and patients with schizophrenia are reviewed. Studies on the association
between specific genes and brain structures and between genes and brain morphological changes in schizophrenia will be reviewed.
A. GENES FOR BRAIN STRUCTURES IN HEALTHY PERSONS
There are eight studies published on the genes for brain structures in healthy
persons. These studies are listed in Table V. These studies looked at brain structures
and specific genes (and alleles or SNPs, single nucleatide polymorphisms) for these
brain structures and at the interaction between these genes.
50
51
Gray matter:
RST gyrus: Z ¼ 3.49, p ¼ 0.000, Thalamus: Z ¼ 3.21, p ¼ 0.001,
RVLPFC: Z ¼ 3.14, p ¼ 0.001, LIPL: Z ¼ 3.08, p ¼ 0.001
White matter:
RVLPFC: Z ¼ 3.68, p ¼ 0.000, LVLPFC: Z ¼ 4.17,
p ¼ 0.000, LHC:
Z ¼ 3.77, p ¼ 0.000
RGS4 (A alleles)
10 MZ twin pairs
(62.5 years)
106 persons
Buckholtz et al. (2007)
Plassman et al. (1997)
McIntosh et al. (2008)
McIntosh et al. (2007)
(2) Wang et al. (2008)
(2) 867 (387 M, 480 F)
75 persons (persons at
risk for SCZ)
87 persons
(1) Woods et al. (2006)
(2) Bueller et al. (2006)
(2) 36 persons
(1) 120 persons
(1) Pezawas et al. (2004)
Study ID
(1) 111 Persons
Sample size
Summary of published studies by August 2009 (PubMed and Medline citations) on genes for brain structural phenotypes in healthy persons. The studies
report on the brain structures that is investigated, the sample size, the results of the intereaction between genes (alleles) and brain morphological changes
(significant or nonsignificant association (n.s.), a decrease (#) or increase (") in brain volume/density in interaction with a gene or an allele) and on the (candidate)
genes/alleles.
BDNF, brain derived neurotrophic factor; COMT, catechol-o-methyl transferase; NRG, neuroglin 1; MCPH1, microcephalin 1; ASPM, abnormal spindlelike microcephaly associated; RGS4, regulator of G protein signaling; C, controls; SCZ, schizophrenia; MZ, monozygotic; DZ, dizygotic; RVLPFC, right
ventrolateral prefrontal cortex; LVLPFC, left ventrolateral prefrontal cortex; LIPL, left inferior parietal lobule; RST gyrus, right superior temporal gyrus.
APOEe4
Neuroglin 1 (rs69994992)
COMT
MCHP1 ASPM
(1) Hippocampal gray matter: Left HC GM: p ¼ 0.013,
t ¼ 2.24, Right: p < 0.001, t ¼ 3.41
(2) Hippocampal volume: # 12,5%, df ¼ 33.1,
F ¼ 4.24, P ¼ 0.048
(1) Brain volume: MCHP1: n.s. (df ¼ 2, 112,
P ¼ 0.20), ASPM: n.s. (df ¼ 2, 114, p ¼ 0.49)
(2) Cranial volume: MCHP1: " volume in males with allele
rs 1057090 (R ¼ 0.126, P ¼ 0.016
Gray matter density in anterior cingulate cortex: #
Gray Matter density
White matter density: # White Matter density ant
limb of the internal capsula
(t ¼ 4.41, p ¼ 0.0028 (T allele). # Fractional anisotropy
in C allele homozygotes
e4 allele: # Hippocampal volume
Association with brain structures
BDNF (both Val66 Met allele)
Genes (alleles)
TABLE V
GENES FOR STRUCTURAL BRAIN PHENOTYPES IN HEALTHY PERSONS
HERITABILITY OF STRUCTURAL BRAIN TRAITS
In some of the studies the interaction of the specific genes with the variance in
brain morphology is expressed in quantitative measures.
As presented in Table V, the genetics of heritability of brain structures indicate
that in one study (Wang et al., 2008) there is a significant association ( p ¼ 0.016)
between the MCPH1 gene and intracranial volume, reaching bigger intracranial
volume in patients with the rs1057090 allele and not reaching a significant
association in another study (Woods et al., 2006), not for the MCPH1 gene
( p ¼ 0.20) nor for the ASPM gene ( p ¼ 0.49). The heritability of gray matter
density has been investigated in three studies. In the first study, there was a
significant association for the right ( p < 0.001) and left hippocampus
( p ¼ 0.013) for the BDNF gene for the Val66met allele carriers (Pezawas et al.,
2004). In the second study, there was a significant association in the right superior
gyrus, temporal lobe, right prefrontal cortex, and left inferior parietal lobe
( p ¼ 0.001) (Buckholtz et al., 2007) for the RGS4 gene and specifically persons
carrying the A allele. In the third study, there was a significant association in the
anterior cingulate cortex for the COMT gene, specifically for persons carrying the
Val gene (McIntosh et al., 2007). The white matter density shows a significant
association with two genes, namely, the RGS4 gene (A allele carriers, p ¼ 0.000) in
the right and left ventral prefrontal cortex and left hippocampus in one study
(Buckholtz et al., 2007) and neuroglin 1 (rs699994992 allele carriers, p ¼ 0.00028)
with a white matter density decrease in anterior limb of the internal capsula in
another study (McIntosh et al., 2007). The hippocampal volume and the association with the APOEe4 gene shows HC volume decrease in one study (Plassman
et al., 1997) and of 11.2% reduction in volume in another study ( p ¼ 0.048)
(Bueller et al., 2006) for the BDNF gene Val66met allele carriers.
The samples consist of healthy individuals, not related, except for one study in
which 10 MZ twin pairs were compared (Plassman et al., 1997) and in another
study, persons at risk for schizophrenia were included (McIntosh et al., 2007). In
the study by Woods et al. (2006), individuals from different ethnic backgrounds
were included, but correcting the data for race, ethnicity or sex did not change the
results. The sample in the study by Wang et al. (2008) consisted of 90% Chinese
and 10% other ethnic minorities. The study by Buckholtz et al. (2007) consisted of
Caucasian subjects of European descent. The sample in the study by McIntosh
et al. (2007) consisted of persons at risk for schizophrenia.
The method used by the studies in Table III varies from one study to the
other. In the study by Woods et al. (2006), heritability of brain volume related to
genotype was estimated by using the coefficient of the genetic terms in the full
fitted ANOVA model to compute the variance attributable to genotype.
This variance was then divided by the total variance to estimate heritability.
This method was also used by Bueller et al. (2006) and Pezawas et al. (2004).
The study by Wang et al. (2008) used the linear regression model to do the single
SNP association analysis. Buckholtz et al. (2007) used multimodal neuroimaging to
52
KAYMAZ AND VAN OS
investigate the impact on brain structure and function and looking at allelic
variation. The study by McIntosh et al. (2007) imported the data into SAS (version
9.1, SAS Institute, Cary, North Carolina) and the effects of group status by
COMT group interactions were investigated by analysis of variance.
B. GENES FOR BRAIN STRUCTURES IN PATIENTS WITH SCHIZOPHRENIA
There are 23 studies where the genetics of variation on heritability of brain
structures was investigated in patients with schizophrenia compared to healthy
controls. These studies report on the association between specific genes and brain
morphological changes in schizophrenia. Several candidate genes for schizophrenia
have been studied and the association with specific brain morphological changes has
been examined. There are multiple studies on the Brain Derived Neurotrophic
Factor gene (BDNF gene) (five studies), Catecol-O-Methyl Transferase gene (COMT
gene) (three studies), Disrupted in Schizophrenia 1 (DISC1) (two studies), APOE e4
gene (two studies) and Interleukin-1 receptor antagonist gene (IL 1) (two studies).
These studies are listed in Table VI. There are single studies on the Neuroglin 1
(NRG1), prion protein (PP gene), tumor necrosis factor receptor-II gene (TNF-RII
gene), NOTCH4 gene, PCM1 gene, Plexin B3 gene, GAD1 gene, PROHD gene,
regulation of G protein signaling (RGS4) and neurotrophin gene. These studies are listed
in Table VII. The brain structures that are looked at in multiple studies include: the
brain lobes (six studies), hippocampal volume (eight studies), ventricles (three studies),
gray matter volume (seven studies), white matter volume (seven studies) and cerebrospinal fluid (CSF) volume (three studies). The following structures have been studied
in a single study: cerebral asymmetry, intracranial volume, regional gray matter
volume/density, prefrontal cortex, caudate, putamen, and cerebellum.
One of the most investigated genes is the BDNF gene. BDNF is a neurotrophic
factor involved in the development and maintenance of the nervous system. It is
neuroprotective and prevents cell loss in the cerebral cortex, the striatum, and the
hippocampus (Angelucci et al., 2005). The BDNF gene consists of several polymorphisms that affect the amino acid sequence and possibly the function of the
protein. In patients with schizophrenia, lower concentrations of BDNF have been
found in cortical areas and the hippocampus as well as in the blood serum.
The COMT gene is thought to be involved in the degradation of dopamine.
The BDNF allele, in particular the Met allele, was found to be associated with a
decrease in hippocampal volume (Bueller et al., 2006) and with frontal lobe and
occipital lobe gray matter in two other studies (Ho et al., 2006; Pezawas et al.,
2004). Szeszko et al. (2005) found an association in a post hoc analysis between
hippocampal volume decrease for schizophrenia patients only and no association
with the intracranial volume. Wassink et al. (1999) studied the BDNF alleles in 48
trios and found an association with the parietal lobe, but not with the temporal or
53
54
(3) Val allele,
PRODH
(rs20086720)
COMT
(1) Val158Met
(2) Polymorphisms
(5) Val66Met
(4) Polymorphisms
(3) Met carriers
BDNF
(1) BDNF
(2) Val66Met vs
val/val
Genes (alleles)
(1) FL, GM, WM, CSF: n.s. (F ¼ < 1.43, df ¼ 2.148, P > 0.24)
(2) Val COMT: ACC (bilateral) p ¼ 0.007, MTG (R) ¼ 0.016, Amygdalauncus (L) p ¼ 0.01, thalamus (left) p ¼ 0.014
(3)"GM density in right superior temporal gyrus (AA genotype of COMT),
p ¼ 0.039,#WM density bilateral frontal lobes (GG genotype of
PRODH), p < 0.05, COMT val allele epistasis with PRODH alleles:
"WM left anterior lobe
(1) Parietal lobe: ¼ 7.90, p ¼ 0.0049, n.s. with FL, TL, OL, ventricles
(2) Hippocampus: F ¼ –3.22, df ¼ 42, p ¼ 0.002 (SCZ þ C); F ¼ 11.72, df
¼ 15, p ¼ 0.004 (SCZ), Intracraniaal volume: t ¼ –1.24, df ¼ 42, p ¼
0.221 (SCZ þ C)
(3) GM volume in 2 cognitive domains: # GM vol in TL: F ¼ 5.44, p ¼ 0.02
and OL: F ¼ 5.73, p ¼ 0.02
(4) All brain structures in literature identified abnormal in SCZ: 111757G/
C: larger frontal GM (p < 0.001) (SCZ), 270 C/T: totale caudate volume
(p ¼ 0.006) and GM in caudate (p ¼ 0.022), 111757G/C þ 270 C/T þ 633T/A þ Val66Met ¼ n.s. for HC, n.s. for controls either with any
brain structure except putamen (p ¼ 0.006)
(5) #Frontal GM in Met allele carriers (F ¼ 4.20, df ¼ 1, 113, p ¼ 0.05),
"lateral ventricles and sulcal CSF (t3.95, df ¼ 44, p0.0003), " CSF in
FL and TL (F6.71, df ¼ 1, 113, p0.01), n.s. in PL and OL (F1.84, df
¼ 1, 113, p0.18)
Association of genes/alleles
with brain structures
(3) 51 SCZ (M)
(1) 100 SCZ, 49 C
(2) 30 SCZ, 104 C
(5) 119 SCZ
(4) 30 SCZ, 104 C
(3) 183 SCZ, 80 C
(1) 48 trios
(2) 19 SCZ, 25 C
Sample size
(continued )
(3) Zinkstok et al. (2007)
(1) Ho et al. (2005)
(2) Ohnishi et al. (2006)
(5) Ho et al. (2007)
(4) Agartz et al. (2006)
(3) Ho et al. (2006)
(1) Wassink et al. (1999)
(2) Szeszko et al. (2005)
Study ID
TABLE VI
MOST STUDIED GENES FOR STRUCTURAL BRAIN PHENOTYPES IN PATIENTS WITH SCHIZOPHRENIA COMPARED TO HEALTHY PERSONS
55
(1) HC: t ¼ 0.23, df ¼ 33, p ¼ 0.82, TL volume: t ¼ 0.27, df ¼ 33, p ¼ 0.79
(2) HC: n.s. (T ¼ 1.9, df ¼ 19, p < 0.10), Cerebral Asymmetry: n.s. (T1.9, df
¼ 19, P < 0.10)
(1) GM volume: Frontal: F ¼ 6.06, df ¼ 1, p < 0.02, Temporal: F ¼ 11.62,
df ¼ 1, p ¼ 0.001, WM volume whole brain: F ¼ 5.54, df ¼ 1, p < 0.03
(2) " Ventricles: Left: t ¼ 3.504, df-21, p ¼ 0.002, Right: t ¼ 2.784, df ¼ 21,
p ¼ 0.01, HC: t < 1.6, df ¼ 21, p ¼ n.s., DLPFC GM: T < 1.5, df ¼ 21,
p ¼ n.s. (p ¼ 0.06)
(1) HEP1 allele: #focal GM density sup þ inf frontal gyri (p < 0.05) (p < 8.61
x10–31 and P < 1.21 10–(8) #HC volume (2 ¼ 3.42, p ¼ 0.06, trend),
HEP2/HEP3 allele: n.s. HC volume change (p ¼ 0.58), AATG
Haplotype:#focal GM density in superior þ middle frontal gyri and
superior temporal gyrus and superior parietal cortex
(2) #HC gray matter volume (Serine allele) (d ¼ 0.36, p < 0.05)
Association of genes/alleles
with brain structures
(1) Meisenzahl et al. (2001)
(2) Papiol et al. (2005)
(2) 23 SCZ, 45 C
(1) Fernandez et al. (1999)
(2) Hata et al. (2002)
(2) Callicott et al. (2005)
(1) Cannon et al. (2005)
Study ID
(1) 44 SCZ, 48 C
(1) 40 SCZ COS, 57 C
(2) 21 SCZ
(2) 86 SCZ (Serine), 72 C
(Cysteine)
(1) 6 MZ þ 1 DZ twins
pairs cc for
SCZ, 28 MZ and
31 DZ C twin pairs
Sample size
Summary of published studies by August 2009 (PubMed and Medline citations) on genes for brain structural phenotypes including study results from
patients with schizophrenia (SCZ) compared to healthy controls (C) in most studies. The studies report on the brain structure that is investigated, the sample
size, the results of the intereaction between genes (alleles/SNPs, single nucleotide polymorphisms) and brain morphological changes (significant or nonsignificant association (n.s.), a decrease (#) or increase (") in brain volume/density in interaction with a gene or an allele or a SNP).
SCZ, schizophrenia; SCZ COS, schizophrenia childhood onset (before age 12); C, controls; cc, concordant and disc, discordant; HC, hippocampus; GM
volume, gray matter volume; WM volume, white matter volume; CSF volume, cerebrospinal fluid spaces volume; DLPFC GM, dorsolateral prefrontal cortex
gray matter; ACC, anterior cingulate cortex; MTG, middle temporal gyrus; TL volume, Temporal lobe volume; BDNF, brain derived neurotrophic factor;
COMT, catechol-o-methyl transferase; RGS4, regulators of G protein signalling; NRG1, neuroglin 1; PP, prion protein gene; NT3 gene, neurotrophin 3 gene;
APOE, apolipoprotein E; PRNP, prion protein; DISC1, disrupted in schizophrenia 1; IL-1RN, interleukin -1 receptor antagonist; TNF, tumor necrosis factor;
NOTCH4 gene; PCM1 gene; PLXN B3, plexin B3 gene; GAD1 gene; ph, polymorphism; L, left; R, right; n.s., not significant; trios, both parents and a child;
vol, volume; ant, anterior; sup, superior; inf, inferior.
IL 1
APOE e4
(2) Serine/
Cysteine allele
DISC1
(1) Rare DISC/
TRAX
haplotypes
Genes (alleles)
TABLE VI (continued )
56
WM volume: F ¼ 5.285, df ¼ 1/84, p ¼ 0/024, CSF volume:
F ¼ 4.400, df ¼ 1/84, p ¼ 0.039, GM volume: F ¼ 1.843,
df ¼ 1/84, p ¼ 0.178 (n.s.)
Frontal lobe (GM, WM, CSF): n.s. ( ¼ 7.09, df ¼ 6, p ¼ 0.31)
GM volume: # GM volume in 6 out of 10 SNPs (p ¼ 0.04–0.003)
DLPFC: # Left DLPFC GM volume, SNP4 (p ¼ 0.0072), SNP18
(p ¼ 0.012), # Right DLFPC GM volume, SNP1 (p ¼ 0.0024),
SNP4 (p ¼ 0.012), SNP7 (p ¼ 0.022)
Gray matter bilateral orbitofrontal: # GM volume, F ¼ 4.6, p < 0.05
Gray matter in temporal lobe, HC, inferior temporal cortex: # GM,
F ¼ 12.7, p < 0.01
GM trajectories, volumes: " (p ¼ 0.001)
WM trajectories, volumes: " (p ¼ 0.04)
WM: F ¼ 4.275, df ¼ 1/85, p ¼ 0.042, GM: F ¼ 2.362, df ¼ 1/85,
p ¼ 0.128, CSF: F ¼ 5.302, df ¼ 1/85, p ¼ 0.024
HC volume: Mean Left HC: t ¼ 1.87, df ¼ 10, p ¼ 0.091, Mean
Right HC: t ¼ 2.72, df ¼ 10, p ¼ 0.022
Allele 1: " ventricle volume, # Frontal lobe volume
Association of genes/alleles with brain structures
Addington et al. (2007)
Rujescu et al. (2007)
78 SCZ COS, 165 C
48 SCZ, 48 C
Gurling et al. (2006)
Wassink et al. (2003)
Addington et al. (2005)
Prasad et al. (2005)
158 SCZ, 48 C
72 SCZ COS, 25 C
30 SCZ, 27 C
14 SCZ (allele þ), 14 C
13 SCZ (allele –), 14 C
Rujescu et al. (2002)
Wassink et al. (2000)
Kunugi et al. (1999)
Study ID
140 probands with SCZ, 197
parents of probands, 46 C
43 SCZ and 47 C
12 SCZ
Sample size
Summary of published studies by August 2009 (PubMed and Medline citations) on genes for brain structural phenotypes including study results from
patients with schizophrenia (SCZ) compared to healthy controls (C) in most studies. The studies report on the brain structure that is investigated, the sample size,
the results of the intereaction between genes (alleles/SNPs, single nucleotide polymorphisms) and brain morphological changes (significant or nonsignificant
association (n.s.), a decrease (#) or increase (") in brain volume/density in interaction with a gene or an allele or a SNP).
SCZ, schizophrenia; SCZ COS, schizophrenia childhood onset (before age 12); C, controls; cc, concordant and disc, discordant; HC, hippocampus; GM
volume, gray matter volume; WM volume, white matter volume; CSF volume, cerebrospinal fluid spaces volume; DLPFC GM, dorsolateral prefrontal cortex
gray matter; ACC, anterior cingulate cortex; MTG, middle temporal gyrus; TL volume, Temporal lobe volume; BDNF, brain derived neurotrophic factor;
COMT, catechol-o-methyl transferase; RGS4, regulators of G protein signalling; NRG1, neuroglin 1; PP, prion protein gene; NT3 gene, neurotrophin 3 gene;
APOE, apolipoprotein E; PRNP, prion protein; DISC1, disrupted in schizophrenia 1; IL-1RN, interleukin -1 receptor antagonist; TNF, tumor necrosis factor;
NOTCH4 gene; PCM1 gene; PLXN B3, plexin B3 gene; GAD1 gene; ph, polymorphism; L, left; R, right; n.s., not significant; trios, both parents and a child;
vol, volume; ant, anterior; sup, superior; inf, inferior.
NRG1 (risk allele 420M9–
1395)
PLXNB3 (haplotype Aþ)
PCM1 (allele þ or – on
chromosome 8)
NOTCH4 (CTG)
GAD1
RGS4 (non G allele)
TNF-RII (homozygous
allele 1)
PRNP gene (Met codon
129 allele)
NT gene (A3 allele)
Genes (alleles)
TABLE VII
GENES FOR STRUCTURAL BRAIN PHENOTYPES IN PATIENTS WITH SCHIZOPHRENIA COMPARED TO HEALTHY CONTROLS, GENES STUDIED IN SINGLE STUDIES
KAYMAZ AND VAN OS
frontal lobe. Ho et al. (2007) found an association between Met allele and a
decrease in frontal gray matter and with an increase in lateral ventricles and in
CSF sulcal volume and an increase in CSF volume in the frontal and temporal
lobe, but not in the parietal and occipital lobe. Agartz et al. (2006) found a
significant association of the 11757 G/C polymorphism with larger frontal gray
matter and a 270C/T ph association with caudate volume and gray matter
volume in caudate and no association with hippocampus, nor with brain volumes
in controls except for the association between the 633 T/A polymorphisms and
putamen. This COMT gene is the second most investigated gene and is located
in a region of chromosome 22q11, which is deleted in patients with the velocardio-facial syndrome. In 25% of these patients, schizophrenia is diagnosed as
well. In a recent metanalyses the COMT gene has not been strongly associated
with schizophrenia (Williams et al., 2007). DISC1 gene is one of the genes found to
be associated with several psychiatric disorders, such as schizophrenia, bipolar
disorder, and depression (Hennah et al., 2006). This gene was studied in a large
Scottish family study, where a translocation was found to be inherited with
different psychiatric disorders (Porteous, 2008). In persons with schizophrenia,
there was a disruption of a gene of unknown function on one of the chromosomal
breakpoints, and was therefore named disrupted in schizophrenia 1. This gene
and the enzymes it codes for are of influence to neuronal migration and memory.
In the study by Cannon et al. (2005), logistic regression was used to evaluate
associations of DISC1 and TRAX haplotypes with diagnostic status, entering age,
sex, substance use disorder, and parental social class as covariates and controlling
for the dependency of observations within twin pairs using generalized estimation
equations. The data of this study show that one of the three haplotypes (HEP 1)
was associated with schizophrenia. The HEP1 haplotype (i.e., TCG) defines a
3-SNP segment located near the translocation break point of DISC1. The HEP1
haplotype was more often associated with individuals with schizophrenia (61.0%;
2 ¼ 6.8 [p ¼ 0.02]; OR ¼ 2.6; 95% CI ¼ 1.2–5.6) when a recessive mode of
transmission was assumed, but not among their nonschizophrenic co-twins
(49.1%; 2 ¼ 0.8 [p ¼ 0.38]). This haplotype was also significantly associated
with small, focal reductions of gray matter in the superior and inferior frontal gyri.
The HEP1 haplotype also shows evidence of an association with hippocampal
volume reduction (2 ¼ 3.42, p ¼ 0.06). The regions affected by DISC1 and
TRAX haplotypes identified in these maps have previously been shown to be
highly heritable in normal twins (Thompson et al., 2001). These regions are also
sensitive to the degree of genetic relationship to an affected individual in twins
discordant for schizophrenia (Cannon et al., 2002; Van Erp et al., 2004).
The APOE e4 allele is a major risk allele for late-onset Alzheimer’s disease
(Mahaney et al., 1993). The study by Hata et al. (2002) shows similar findings to
studies carried out in patients with Alzheimer’s disease (Soininen and Riekkinen,
1996; Tohgi et al., 1997) in that there is a decrease in hippocampal volume of the
57
HERITABILITY OF STRUCTURAL BRAIN TRAITS
e4 allele carriers compared to noncarriers. The IL-RII gene has been studied
twice until now, one showing an association between a high activity allele in the
interleukin-1b (IL-1b) gene and orbitofrontotemporal volume reductions of gray
and white matter in schizophrenia subjects but not in controls (Meisenzahl et al.,
2001) and the other study showing variability at the IL-1RN gene contributing to
the ventricular volumetric variation in patients with schizophrenia. The PP gene,
with a Val/Met polymorphism at position 129 in the prion protein gene (PRNP), is
known as a risk factor for Creutzfeldt-Jakob disease, which is accompanied by
psychosis in the early stages. It is unclear whether this PP gene is also associated
with schizophrenia (Martorell et al., 2007), but in the study by Rujescu et al. (2002),
there was a decrease in white matter volume and larger CSF volume in both the
controls as well as the patients with the Met homozygotes. The other genes are
studied in just one study. These studies have all been analyzed in detail in previous
reviews (Pearlson and Calhoun, 2007; Van Haren et al., 2008).
While the reviewed studies suggest high heritability rates for some brain
structures, limitations on the neuroimaging, twin studies and NHP studies make
the interpretation and the generalizability of their findings difficult.
IV. Limitations and Clinical Relevance of the Studies
A. LIMITATIONS OF NEUROIMAGING STUDIES
A number of limitations in imaging studies need to be acknowledged, such as
the validity and reliability of measures (e.g., using measurements that rely on a
representative slice vs. whole volume or do not correct for total brain volume).
Measurements of brain volumes reveal differences between affected subjects and
healthy controls, but the magnitude of these differences is moderate and there is
always a substantial overlap in the distributions of the comparison groups.
However, looking at the heritability of brain volumes and shapes in healthy
twin studies shows high heritability rates of these brain measurements. One
solution to this problem might be to shape analysis to quantify subtle differences
in brain structures, in order to enhance the identification of structural brain
abnormalities that might be associated with neurological and psychiatric disorders such as schizophrenia. Another limitation of the reported studies is the
sample size, which is a major issue since imaging techniques and genetic studies
require considerable statistical power. Although the gross differences in size or
symmetry of brain structures might be quantified, the individual cells and cell
layers cannot yet be visualized. This means that, although the volume and shape
of brain structures may be determined, the underlying cause of any differences
cannot. Another limitation in imaging studies is the method applied to measure
58
KAYMAZ AND VAN OS
the volumes of brain structures. Some studies use the ROI analyses, where specific
brain areas are manually traced. This method does not accommodate changes at
sites distant to the ROI, and thus may be inherently biased in spatial terms.
Investigating covariation between structures is important, for it may suggest that
different brain structures are being influenced by the same gene or sets of genes.
This approach may also be less sensitive to small variations in brain structure or
tissue type distribution than more modern modeling methods (Good et al., 2001a,
2002). There is also a wide intra- and inter-rater variability, and ROI measurement errors are likely to be greater for smaller structures with complex architectures, particularly if anatomical borders are defined by arbitrary criteria. Other
studies used the VBM method to measure volumes of brain structures, a fully
automated whole brain technique. The VBM method avoids many of the constraints of ROI analyses, but it incorporates a series of preprocessing steps that
may complicate the simple interpretation of regional changes of gray matter loss.
One new method used in the heritability estimates is the elastic deformation
procedure, used by Thompson et al. (2001), in which researchers modeled the
genetic influences on neuroanatomic variation between healthy MZ and DZ twin
pairs at each voxel on the surface of the cortex. Findings from these analyses
showed a significant influence of the genetic factors on the gray matter density
in subregions of prefrontal and temporal lobes, particularly in Broca’s and
Wernicke’s language areas. Other areas within these structures were found less
heritable, suggesting that heritability estimates based on gross lobular regions of
interest blur more subtle local differences.
B. LIMITATIONS OF TWIN STUDIES
Twin studies cannot solve the problem of sources of variance, such as the
gene–environment interactions, nonrandom assortative mating, and the possibility
of differential interactions of environment in MZ and DZ twin pairs and the
chorion type in MZ twins. Reliable detection of differences in correlations may
therefore require hundreds of twin pairs. Interpretation of all the data in the
reported studies is based on three assumptions in the classical twin designs: (1) MZ
twins share 100% and DZ twins share on average 50% of their polymorphic
genetic material and current methods adequately identify zygosity, (2) the environment shared among MZ and DZ twins is similar, and (3) twins are similar to
singletons such that findings in twins can be generalized to nontwin populations.
Conflicting results exist on the first assumption in schizophrenia, specifically that
MZ twins discordant for schizophrenia share the same genetic material. Some
studies (Guidry et al., 1999; Petronis and Kennedy, 1995; Petronis et al., 2003;
Singh et al., 2002) suggest that twins discordant for schizophrenia share less of
their polymorphic material than concordant and healthy twins in contrast to
59
HERITABILITY OF STRUCTURAL BRAIN TRAITS
other studies claiming there is no empirical evidence for this (Nguyen et al., 2003;
Tsujita et al., 1998; Vincent et al., 1998). The second assumption, that twins share
the same environmental factors, has been criticized because of the physical
maternal stress, which may be higher for twins then for singletons. There may
also be more competition between the twins regarding nutrition. Although classic
twin studies attempt to estimate unique environmental effects (events occurring to
one twin but not to the other), as well as genetic (heritability) and environmental
effects, one limitation of the model is that the gene–environment correlation and
interaction is not discernible as a distinct effect. Another limitation is that there is no
opportunity in the classic twin design to consider both shared environment and
nonadditive genetic effects simultaneously. This limitation can be addressed by
including additional siblings to the design, which was done by some studies in the
healthy twin studies (Tables I and II). None of the studies reported in this review
indicate the impact of possible prenatal and perinatal obstetric complications,
except for the study by Van Erp et al. (2004) in which the obstetric records of half
of the studied twin pairs did not differ between discordant MZ and DZ twin pairs,
and the study by Wallace et al. (2006) in which data were corrected for the 29 weeks
postgestational weight of the individuals included in the study. The third assumption on the generalizability of twin data to singletons is not possible according to
some, because twins are not a random sample of the population and they differ in
their developmental environment; in this sense they are not representative
(Kempthorne and Osborne, 1961). Other studies claim that twins differ very little
from their non-twin siblings and that measured studies on personality and intelligence of twins suggests that they have scores on these traits very similar to those of
nontwins (Deary et al., 2005). Only one study, by Hulshoff Pol et al. (2002), addresses
this question and suggests that the data from twin studies can be generalized to
singletons, when correcting for the intracranial volume for the second born twin
compared to the first born, since second born twins have lower intracranial volume.
All other volumes are comparable as well, suggesting that twin studies provide
reliable estimates of heritability and can be generalized to singletons.
C. LIMITATIONS OF NONHUMAN PRIMATE STUDIES
Due to the close anatomical, physiological, biochemical, and genetic similarity
to humans, NHPs play an important role as model in many aspects of research,
specifically for genetic analysis of individual variation in brain structure. The most
used species are the baboons, belonging to the old-world monkeys the rhesus
macaques, cynomolgus macaques, and vervet monkeys, and being very closely
related to the humans. Other primate species that are more similar to humans are
the chimpanzees and gorillas. These species would represent a better model for
the human neuroanatomy, but they present only minimal opportunities
60
KAYMAZ AND VAN OS
for intraspecies genetic analyses due to the lack of extensive multigeneration
pedigrees (Rogers et al., 2007). Due to the use of large colonies, the analysis of
complex phenotypes in the old world monkey can help us understand more about
the genetic basis of individual variation in important traits.
One of the limitations of the NHP studies is that the heritability estimates in
monkeys are specific to the population and circumstances in which they are
assessed. Genetically diverse populations in homogeneous environments demonstrate larger heritability than do inbred populations in diverse environments
(Falconer and Mackay, 1996).
Very few parents in the monkey colonies that make up study cohorts share a
common mother or father, and extended family pedigrees cannot be determined
from monkey breeding colony records. The generality of the NHP studies should
therefore be tested in studies of other populations (Lyons et al., 2001).
It may be clear that the human brain is more than simply a large monkey brain.
On the one hand, the monkey and the human brains share a common plan of
organization, but on the other hand, striking species differences and specific regional
differences exist. In the course of evolution, besides an increase in mean size, the
brain of the newborn human is exposed to the physical, social, and cultural environmental factors for an extended period compared to the monkey, resulting in ‘‘cultural
imprints,’’ such as learning the spoken and written language (Deheane et al., 2005).
D. CLINICAL RELEVANCE OF THE HERITABILITY ESTIMATIONS
Study of the influence of genetic factors on brain structures will help us to
understand normal development and age, gender, and other variable associated
changes, and will moreover help us understand the meaning of morphological
changes found in schizophrenia.
One of the functional relevancies of the heritability of brain volumes is
demonstrated best by the association between brain volumes and intelligence.
In a study by Posthuma et al. (2002b), the association between brain volumes and
intelligence was found to be of genetic origin. In two other studies (Thompson
et al., 2001; Thompson et al., 2002), the association between frontal gray matter
volume and intelligence was shown to be due to genetic factors. In a recent study
by Hulshoff Pol et al. (2006), an association between intelligence and gray matter
of the frontal and occipital lobe, the parahippocampus and the connecting white
matter was found. While these findings suggest a common set of genes and
intelligence, a study by Carmelli et al. (1998), shows that in elderly twins, the
association between frontotemporal brain volumes, and executive function were
found to be associated with common environmental influences shared by twins
from the same family. Whether there is a gene–environment interaction, for
example, persons with high level of cognitive functioning leading to select an
61
HERITABILITY OF STRUCTURAL BRAIN TRAITS
environment that also increases brain size, or the possibility of overlapping sets of
genes or common environmental influences causing variation in two distinct
phenotypes, the cause of the findings between intelligence, and these brain
structures, remains unknown. The heritability estimates are necessary to estimate
the degree to which genetic factors and environmental factors have influence on
the phenotype. Estimation of the impact of environmental factors is much more
complicated and generally the effect size of environmental factors are considered
to be the remaining part of the total heritability of a disease once the heritability
due to genetic factors is known. In addition to the heritability measurements,
genes involved in brain structures that are under the influence of genetic factors
would be more easily investigated than looking at genes for brain structures in
patients with schizophrenia, without knowing the origin (e.g., genetic vs. environmental, a gene–environment interaction or disease related) of these brain changes.
This is where the studies in NHPs might be useful, specifically since the impact of
the environmental factors can be controlled and questions regarding the evolution
of different brain parts and of human cognition only can be addressed adequately if
the differences and similarities in the underlying neural circuitry of humans and all
extant apes can be identified. Some data from older studies have been used by
investigators to address these issues, but problems regarding the methodology used
in the measurements, the incomplete representation of hominoid species and the
small samples included makes the interpretation of these data difficult.
E. IMPLICATIONS OF THESE OF BRAIN VOLUMES AS ENDOPHENOTYPES
FOR GENETICS OF PSYCHIATRIC DISORDERS
Despite the limitations of the twin studies and neuroimaging studies, it is fair
to say that these studies only represent a subset of the possible genetic and
environmental relationship and that the heritability estimates from these studies
represent a first step in understanding the genetics of brain structures.
Using the brain structures with high heritability scores as endophenotypes and
looking at the gene action for these endophenotypes, might solve problems in
neuroimaging and genetic studies. In this study we found that global structures,
with high heritability rates, such as total brain volume, hemispheres, cerebrum,
cerebellum, gray and white matter, the four lobes (frontal, temporal, occipital and
parietal), and corpus callosum seem to be more useful as endophenotype markers
than specific structures with lower heritability rates. These structures are indeed
formed earlier in development and the high heritability rates are suggestive of
more genetic influences on early developed structures. Lower heritability rates for
the specific structures might be due to lower sample sizes found in most of the
studies, due to differences in methods measuring the volumes or due to difference
in age in participating persons. The studies in NHPs show the same results on the
62
KAYMAZ AND VAN OS
heritability of gross brain structures, although there are few studies on the
heritability of brain structures in NHPs. Looking at the genes for brain structures
in healthy persons and persons with schizophrenia shows us that gene approach
studies do not show a causal pathway for the brain morphological changes in
patients with schizophrenia, and no susceptibility gene with a high effect size has
yet been found for schizophrenia.
To summarize, the findings of the reviewed studies suggest that the overall
brain volume is highly heritable, comparable with twin based estimates for the
most highly genetically determined human traits, including fingerprint ridge
count (h2 ¼ 0.98) (Spence et al., 1977), height (h2 ¼ 0.66) (Lettre, 2009;
Silventoinen, 2003), and systolic blood pressure (h2 ¼ 0.57) (Williams et al.,
1990). The structures that are formed earlier in development (Rubenstein and
Rakic, 1999; Rubenstein et al., 1999) and located deeper in the brain may be more
genetically influenced thus, they show higher heritability scores than the structures which are formed later in development (Rakic, 1988) or are located on
surface areas (Nadarajah and Parnavelas, 2002; Rakic, 1972). These structures
are also under more environmental influence. Heritability studies on the genetic
control of the human brain structure and how much individual genotype accounts
for the wide variations among individual brains using neuranatomically defined
endophenotypes will make it easier to look for the actions of genes on brain
structures. However, a critical look at the genetic and or disease related changes in
the brains of patients with complex disorders such as schizophrenia is necessary,
as is suggested by Cannon et al. (2002). Further research on the influences of
genetic, common, and unique environmental factors is necessary with advanced
quantitative genetic methods. Genotyping studies in search of genes for brain
structures in healthy subjects and for brain morphological changes in diseased
need to be replicated; however, until now individual studies have been generally
underpowered in replicating the small number of promising susceptibility genes
(Ross et al., 2006) with small effect sizes (OR < 1.5). Looking for an association
between brain morphological changes with genes, in particular the search for rare
variants for overall genetic risk, with a more advanced method like the Genome
Wide Association studies in combination with brain morphological changes,
could bring new findings in causal pathways wherein genes interact with brain
morphology. This may also shed light on understanding the relationship between
schizophrenia and other major psychiatric disorders.
Acknowledgments
This research was supported by the Netherlands Organization for Scientific Research (NWO) under
project number 017.002.048. We also thank A. Jasinska, Ph.D., UCLA/Los Angeles/USA, for her advices.
63
HERITABILITY OF STRUCTURAL BRAIN TRAITS
References
Addington, A. M., Gornick, M., Duckworth, J., Sporn, A., Gogtay, N., Bobb, A., Greenstein, D.,
Lenane, M., Gochman, P., Baker, N., Balkissoon, R., Vakkalanka, R. K., et al. (2005). GAD1
(2q31.1), which encodes glutamic acid decarboxylase (GAD67), is associated with childhood-onset
schizophrenia and cortical gray matter volume loss. Mol. Psychiatry 10(6), 581–588.
Addington, A. M., Gornick, M. C., Shaw, P., Seal, J., Gogtay, N., Greenstein, D., Clasen, L.,
Coffey, M., Gochman, P., Long, R., and Rapoport, J. L. (2007). Neuregulin 1 (8p12) and
childhood-onset schizophrenia: Susceptibility haplotypes for diagnosis and brain developmental
trajectories. Mol. Psychiatry 12(2), 195–205.
Agartz, I., Sedvall, G. C., Terenius, L., Kulle, B., Frigessi, A., Hall, H., and Jönsson, E. G. (2006).
BDNF gene variants and brain morphology in schizophrenia. Am. J. Med. Genet. B Neuropsychiatr.
Genet 141B(5), 513–523.
Alexander, M. P., and Annett, M. (1996). Crossed aphasia and related anomalies of cerebral organization: Case reports and a genetic hypothesis. Brain Lang. 55(2), 213–239. Review.
Allen, J. S., Damasio, H., and Grabowski, T. J. (2002). Normal neuroanatomical variation in the
human brain: An MRI-volumetric study. Am. J. Phys. Anthropol. 118(4), 341–358.
Angelucci, F., Brenè, S., and Mathé, A. A. (2005). BDNF in schizophrenia, depression and
corresponding animal models. Mol. Psychiatry 10(4), 345–352. Review.
Annett, M. (1974). Handedness in the children of two left-handed parents. Br. J. Psychol. 65(1), 129–131.
Ashburner, J., and Friston, K. J. (2000). Voxel-based morphometry: The methods. Neuroimage 11
(6 Pt 1), 805–821. Review.
Atwood, L. D., Wolf, P. A., Heard-Costa, N. L., Massaro, J. M., Beiser, A., D’Agostino, R. B., and
DeCarli, C. (2004). Genetic variation in white matter hyperintensity volume in the Framingham
Study. Stroke 35(7), 1609–1613.
Baaré, W. F., Hulshoff Pol, H. E., Boomsma, D. I., Posthuma, D., de Geus, E. J., Schnack, H. G., van
Haren, N. E., van Oel, C. J., and Kahn, R. S. (2001a). Quantitative genetic modeling of variation
in human brain morphology. Cereb Cortex 11(9), 816–824.
Baaré, W. F., van Oel, C. J., Hulshoff Pol, H. E., Schnack, H. G., Durston, S., Sitskoorn, M. M., and
Kahn, R. S. (2001b). Volumes of brain structures in twins discordant for schizophrenia. Arch. Gen.
Psychiatry 58(1), 33–40.
Bartley, A. J., Jones, D. W., and Weinberger, D. R. (1997). Genetic variability of human brain size and
cortical gyral patterns. Brain 120((Pt 2), 257–269.
Bentall, R. P., Read, J. E., and Mosher, L. R. (2004). Models of Madness: Psychological, Social and
Biological Approaches to Schizophrenia Brunner-Routledge, Philadelphia. ISBN 1-58391-906-6.
Biondi, A., Nogueira, H., Dormont, D., Duyme, M., Hasboun, D., Zouaoui, A., Chantôme, M., and
Marsault, C. (1998). Are the brains of monozygotic twins similar? A three-dimensional MR study.
AJNR Am. J. Neuroradiol. 19(7), 1361–1367.
Blangero, J., Williams, J. T., and Almasy, L. (2003). Novel family-based approaches to genetic risk in
thrombosis. J. Thromb. Haemost 1(7), 1391–1397.
Boyle, M. (2002). Schizophrenia: A Scientific Delusion? Routledge, New York. ISBN 0-415-22718-6.
Boos, H. B., Aleman, A., Cahn, W., Hulshoff Pol, H., and Kahn, R. S. (2007). Brain volumes in
relatives of patients with schizophrenia: A meta-analysis. Arch. Gen. Psychiatry 64(3), 297–304.
Boomsma, D., Busjahn, A., and Peltonen, L. (2002). Classical twin studies and beyond. Nat. Rev. Genet.
3(11), 872–882. Review.
Brans, R. G., van Haren, N. E., van Baal, G. C., van Baal, G. C., Schnack, H. G., Kahn, R. S., and
Hulshoff Pol, H. E. (2008). Heritability of changes in brain volume over time in twin pairs
discordant for schizophrenia. Arch. Gen. Psychiatry 65(11), 1259–1268.
64
KAYMAZ AND VAN OS
Bremner, J. D., Krystal, J. H., Southwick, S. M., and Charney, D. S. (1995). Functional neuroanatomical
correlates of the effects of stress on memory. J. Trauma Stress 8(4), 527–553. Review.
Bremner, J. D., Narayan, M., Anderson ER Staib, L. H., Miller, H. L., and Charney, D. S. (2000).
Hippocampal volume reduction in major depression. Am. J. Psychiatry 157(1), 115–118.
Bruckner, C. M., and Slanger, W. D. (1986). Symmetric differences squared and analysis of variance
procedures for estimating genetic and environmental variances and covariances for beef cattle
weaning weight: (a) Comparison via simulation (b) Estimates from a data set. J. Anim. Sci. 63(6),
1779–1793.
Buckholtz, J. W., Meyer-Lindenberg, A., Honea, R. A., Straub, R. E., Pezawas, L., Egan, M. F.,
Vakkalanka, R., Kolachana, B., Verchinski, B. A., Sust, S., Mattay, V. S., Weinberger, D. R., et al.
(2007). Allelic variation in RGS4 impacts functional and structural connectivity in the human
brain. J. Neurosci. 27(7), 1584–1593.
Bueller, J. A., Aftab, M., Sen, S., Gomez-Hassan, D., Burmeister, M., and Zubieta, J. K. (2006). BDNF
Val66Met allele is associated with reduced hippocampal volume in healthy subjects. Biol. Psychiatry
59(9), 812–815.
Cahn, W., Hulshoff Pol, H. E., Bongers, M., Schnack, H. G., Mandl, R. C., Van Haren, N. E.,
Durston, S., Koning, H., Van Der Linden, J. A., and Kahn, R. S. (2001). Brain morphology in
antipsychotic-naı̈ve schizophrenia: A study of multiple brain structures. Br. J. Psychiatry. 43
(Suppl.), s66–s72.
Callicott, J. H., Straub, R. E., Pezawas, L., Egan, M. F., Mattay, V. S., Hariri, A. R., Verchinski, B. A.,
Meyer-Lindenberg, A., Balkissoon, R., Kolachana, B., Goldberg, T. E., and Weinberger, D. R.
(2005). Variation in DISC1 affects hippocampal structure and function and increases risk for
schizophrenia. Proc. Natl Acad. Sci. USA 102(24), 8627–8632.
Cardno, A. G., and Gottesman, A. G. (2000). Twin studies of schizophrenia: From bow-and-arrow
concordances to star wars Mx and functional genomics. Am. J. Med. Genet 97(1), 12–17. Review.
Cardno, A. G., Thomas, K., and McGuffin, P. (2002). Clinical variables and genetic loading for
schizophrenia: Analysis of published Danish adoption study data. Schizophr. Bull. 28(3), 393–399.
Carmelli, D., DeCarli, C., Swan, G. E., Jack, L. M., Reed, T., Wolf, P. A., and Miller, B. L. (1998).
Evidence for genetic variance in white matter hyperintensity volume in normal elderly male twins.
Stroke 29(6), 1177–1181.
Cannon, T. D., Kaprio, J., Lönnqvist, J., Huttunen, M., and Koskenvuo, M. (1998). The genetic
epidemiology of schizophrenia in a Finnish twin cohort. A population-based modeling study. Arch.
Gen. Psychiatry 55(1), 67–74.
Cannon, T. D., Thompson, P. M., van Erp, T. G., Toga, A. W., Poutanen, V. P., Huttunen, M.,
Lonnqvist, J., Standerskjold-Nordenstam, C. G., Narr, K. L., Khaledy, M., Zoumalan, C. I.,
Dail, R., et al. (2002). Cortex mapping reveals regionally specific patterns of genetic and diseasespecific gray-matter deficits in twins discordant for schizophrenia. Proc. Natl Acad. Sci. USA 99(5),
3228–3233.
Cannon, T. D., van Erp, T. G., Bearden, C. E., Loewy, R., Thompson, P., Toga, A. W.,
Huttunen, M. O., Keshavan, M. S., Seidman, L. J., and Tsuang, M. T. (2003). Early and late
neurodevelopmental influences in the prodrome to schizophrenia: Contributions of genes, environment, and their interactions. Schizophr. Bull. 29(4), 653–669.
Cannon, T. D., Hennah, W., van Erp, T. G., Thompson, P. M., Lonnqvist, J., Huttunen, M.,
Gasperoni, T., Tuulio-Henriksson, A., Pirkola, T., Toga, A. W., Kaprio, J., Mazziotta, J., et al.
(2005). Association of DISC1/TRAX haplotypes with schizophrenia, reduced prefrontal gray
matter, and impaired short- and long-term memory. Arch. Gen. Psychiatry 62(11), 1205–1213.
Cantalupo, C., and Hopkins, W. D. (2001). Asymmetric Broca’s area in great apes. Nature 414(6863), 505.
Carne, R. P., Vogrin, S., Litewka, L., and Cook, M. J. (2006). Cerebral cortex: An MRI-based study of
volume and variance with age and sex. J. Clin. Neurosci 13(1), 60–72.
65
HERITABILITY OF STRUCTURAL BRAIN TRAITS
Cheverud, J. M., Falk, D., Vannier, M., Moore, A. J., Helmkamp, R. C., and Vannier, M. (1990).
Heritability of brain size and surface features in rhesus macaques (Macaca mulatta). J. Hered 81(1),
51–57.
Cosgrove, K. P., Mazure, C. M., and Staley, J. K. (2007). Evolving knowledge of sex differences in
brain structure, function, and chemistry. Biol. Psychiatry 62(8), 847–855. Review.
Deary, I. J., Pattie, A., Wilson, V., and Whalley, L. J. (2005). The cognitive cost of being a twin: Two
whole-population surveys. Twin. Res. Hum. Genet 8(4), 376–383.
Dehaene, S., Duhamel, J.-R. D., Hauser, M. D., and Rizzolatti, G. (2005). Genes, brains, and culture:
From monkey to human. In ‘‘Monkey Brain to Human Brain’’.
Deutch, A. Y. (1993). Prefrontal cortical dopamine systems and the elaboration of functional corticostriatal circuits: Implications for schizophrenia and Parkinson’s disease. J. Neural Transm. Gen. Sect.
91(2–3), 197–221. Review.
Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-Tr) (2000). American Psychiatric
Association (APA).
Driemeyer, J., Boyke, J., Gaser, C., Büchel, C., and May, A. (2008). Changes in gray matter induced by
learning-revisited. PLoS One 3(7), e2669.
Eckert, M. A., Leonard, C. M., Molloy, E. A., Blumenthal, J., Zijdenbos, A., and Giedd, J. N. (2002).
The epigenesis of planum temporale asymmetry in twins. Cereb. Cortex 12(7), 749–755.
Falconer, D. S., and Mackay, T. F. C. (1996). Introduction to Quantitative Genetics 4th Ed. Addison
Wesley Longman, Harlow, Essex, UK.
Falconer, D. S., Gauld, I. K., Roberts, R. C., and Williams, D. A. (1981). The control of body size in
mouse chimaeras. Genet. Res 38(1), 25–46.
Fears, S. C., Melega, W. P., Service, S. K., Lee, C., Chen, K., Tu, Z., Jorgensen, M. J.,
Fairbanks, L. A., Cantor, R. M., Freimer, N. B., and Woods, R. P. (2009). Identifying heritable
brain phenotypes in an extended pedigree of vervet monkeys. J. Neurosci. 29(9), 2867–2875.
Fernandez, T., Yan, W. L., Hamburger, S., Rapoport, J. L., Saunders, A. M., Schapiro, M.,
Ginns, E. I., and Sidransky, E. (1999). Apolipoprotein E alleles in childhood-onset schizophrenia.
Am. J. Med. Genet 88(2), 211–213.
Gannon, P. J., Holloway, R. L., Broadfield, D. C., and Braun, A. R. (1998). Asymmetry of chimpanzee
planum temporale: Humanlike pattern of Wernicke’s brain language area homolog. Science 279
(5348), 220–222.
Geschwind, N., and Galaburda, A. M. (1985). Cerebral lateralization. Biological mechanisms, associations, and pathology: III. A hypothesis and a program for research. Arch. Neurol. 42(7), 634–654.
Geschwind, D. H., Miller, B. L., DeCarli, C., and Carmelli, D. (2002). Heritability of lobar brain
volumes in twins supports genetic models of cerebral laterality and handedness. Proc. Natl. Acad.
Sci. USA 99(5), 3176–3181.
Giedd, J. N. (2008). The teen brain: Insights from neuroimaging. J. Adolesc. Health 42(4), 335–343. Review.
Good, C. D., Ashburner, J., and Frackowiak, R. S. (2001a). Computational neuroanatomy: New
perspectives for neuroradiology. Rev. Neurol. (Paris) 157(8–9 Pt 1), 797–806. Review.
Good, C. D., Johnsrude, I. S., Ashburner, J., Henson, R. N., Friston, K. J., and Frackowiak, R. S.
(2001b). A voxel-based morphometric study of ageing in 465 normal adult human brains. Neuroimage 14(1 Pt 1), 21–36.
Good, C. D., Scahill, R. I., Fox, N. C., Ashburner, J., Friston, K. J., Chan, D., Crum, W. R.,
Rossor, M. N., and Frackowiak, R. S. (2002). Automatic differentiation of anatomical patterns
in the human brain: Validation with studies of degenerative dementias. Neuroimage 17(1), 29–46.
Goldman, M. B., Torres, I. J., Keedy, S., Marlow-O’Connor, M., Beenken, B., and Pilla, R. (2007).
Reduced anterior hippocampal formation volume in hyponatremic schizophrenic patients.
Hippocampus 17(7), 554–562.
Gottesman, I. I., and Gould, T. D. (2003). The endophenotype concept in psychiatry: Etymology and
strategic intentions. Am. J. Psychiatry 160(4), 636–645. Review.
66
KAYMAZ AND VAN OS
Gould, E., and Tanapat, P. (1999). Stress and hippocampal neurogenesis. Biol. Psychiatry 46(11),
1472–1479. Review.
Guidry, J., and Kent, T. A. (1999). New genetic hypothesis of schizophrenia. Med. Hypotheses. 52(1),
69–75.
Gurling, H. M., Critchley, H., Datta, S. R., McQuillin, A., Blaveri, E., Thirumalai, S., Pimm, J.,
Krasucki, R., Kalsi, G., Quested, D., Lawrence, J., Bass, N., et al. (2006). Genetic association and
brain morphology studies and the chromosome 8p22 pericentriolar material 1 (PCM1) gene in
susceptibility to schizophrenia. Arch. Gen. Psychiatry 63(8), 844–854. Erratum in: Arch. Gen. Psychiatry
64(11), 1258.
Gurvits, T. V., Shenton, M. E., Hokama, H., Ohta, H., Lasko, N. B., Gilbertson, M. W., Orr, S. P.,
Kikinis, R., Jolesz, F. A., McCarley, R. W., and Pitman, R. K. (1996). Magnetic resonance
imaging study of hippocampal volume in chronic, combat-related posttraumatic stress disorder.
Biol. Psychiatry 40(11), 1091–1099.
Hata, T., Kunugi, H., Nanko, S., Fukuda, R., and Kaminaga, T. (2002). Possible effect of the APOE
epsilon 4 allele on the hippocampal volume and asymmetry in schizophrenia. Am. J. Med. Genet
114(6), 641–642.
Hennah, W., Thomson, P., Peltonen, L., and Porteous, D. (2006). Genes and schizophrenia: Beyond
schizophrenia—The role of DISC1 in major mental illness. Schizophr. Bull. 32(3), 409–416. Review.
Ho, B. C., Wassink, T. H., O’Leary, D. S., Sheffield, V. C., and Andreasen, N. C. (2005). Catechol-Omethyl transferase Val158Met gene polymorphism in schizophrenia: Working memory, frontal
lobe MRI morphology and frontal cerebral blood flow. Mol. Psychiatry 10(3), 229, 287–298.
Ho, B. C., Milev, P., O’Leary, D. S., Librant, A., Andreasen, N. C., and Wassink, T. H. (2006).
Cognitive and magnetic resonance imaging brain morphometric correlates of brain-derived
neurotrophic factor Val66Met gene polymorphism in patients with schizophrenia and healthy
volunteers. Arch. Gen. Psychiatry 63(7), 731–740.
Ho, B. C., Andreasen, N. C., Dawson, J. D., and Wassink, T. H. (2007). Association between brainderived neurotrophic factor Val66Met gene polymorphism and progressive brain volume changes
in schizophrenia. Am. J. Psychiatry 164(12), 1890–1899.
Hulshoff Pol, H. E., Posthuma, D., Baaré, W. F., De Geus, E. J., Schnack, H. G., van Haren, N. E., van
Oel, C. J., Kahn, R. S., and Boomsma, D. I. (2002). Twin-singleton differences in brain structure
using structural equation modelling. Brain. 125(Pt 2), 384–390.
Hulshoff Pol, H. E., Brans, R. G., van Haren, N. E., Schnack, H. G., Langen, M., Baaré, W. F., van
Oel, C. J., and Kahn, R. S. (2004). Gray and white matter volume abnormalities in monozygotic
and same-gender dizygotic twins discordant for schizophrenia. Biol. Psychiatry 55(2), 126–130.
Hulshoff Pol, H. E., Schnack, H. G., Posthuma, D., Mandl, R. C., Baaré, W. F., van Oel, C., van
Haren, N. E., Collins, D. L., Evans, A. C., Amunts, K., Bürgel, U., Zilles, K., et al. (2006). Genetic
contributions to human brain morphology and intelligence. J. Neurosci. 26(40), 10235–10242.
Ilg, R., Wohlschläger, A. M., Gaser, C., Liebau, Y., Dauner, R., Wöller, A., Zimmer, C., Zihl, J., and
Mühlau, M. (2008). Gray matter increase induced by practice correlates with task-specific
activation: A combined functional and morphometric magnetic resonance imaging study.
J. Neurosci. 28(16), 4210–4215.
Kanaan, R., Barker, G., Brammer, M., Giampietro, V., Shergill, S., Woolley, J., Picchioni, M.,
Toulopoulou, T., and McGuire, P. (2009). White matter microstructure in schizophrenia: Effects
of disorder, duration and medication. Br. J. Psychiatry 194(3), 236–242.
Kappelman, J. (1993). The evolution of body mass and relative brain size in fossil hominoids. J. Human
Evol. 30, 243–276.
Kempthorne, O., and Osborne, R. H. (1961). The interpretation of twin data. Am. J. Hum. Genet. 13,
320–339.
Kendell, R., and Jablensky, A. (2003). Distinguishing between the validity and utility of psychiatric
diagnoses. Am. J. Psychiatry 160(1), 4–12.
67
HERITABILITY OF STRUCTURAL BRAIN TRAITS
Kendler, K. S., and Diehl, S. R. (1993). The genetics of schizophrenia: A current, genetic-epidemiologic perspective. Schizophr. Bull. 19(2), 261–285.
Koolschijn, P. C., van Haren, N. E., Hulshoff Pol, H. E., and Kahn, R. S. (2007). Hypothalamus
volume in twin pairs discordant for schizophrenia. Eur. Neuropsychopharmacol. 18(4), 312–315.
Kunugi, H., Hattori, M., Nanko, S., Fujii, K., Kato, T., and Nanko, S. (1999). Dinucleotide repeat
polymorphism in the neurotrophin-3 gene and hippocampal volume in psychoses. Schizophr. Res.
37(3), 271–273.
Lawrie, S. M., and Abukmeil, S. S. (1998). Brain abnormality in schizophrenia. A systematic and
quantitative review of volumetric magnetic resonance imaging studies. Br. J. Psychiatry 172,
110–120.
Lee, H., Devlin, J. T., Shakeshaft, C., Stewart, L. H., Brennan, A., Glensman, J., Pitcher, K.,
Crinion, J., Mechelli, A., Frackowiak, R. S., Green, D. W., and Price, C. J. (2007). Anatomical
traces of vocabulary acquisition in the adolescent brain. J. Neurosci. 27(5), 1184–1189.
Lettre, G. (2009). Genetic regulation of adult stature. Curr. Opin. Pediatr. 21(4), 515–522.
Linscott, R. J., Allardyce, J., and van Os, J. (2009). Seeking verisimilitude in a class: A systematic review
of evidence that the criterial clinical symptoms of schizophrenia are taxonic. Schizophr. Bull.
[Epub ahead of print].
Lohmann, G., von Cramon, D. Y., and Steinmetz, H. (1999). Sulcal variability of twins. Cereb. Cortex
9(7), 754–763.
Lyons, D. M., Yang, C., Sawyer-Glover, A. M., Moseley, M. E., and Schatzberg, A. F. (2001). Early life stress
and inherited variation in monkey hippocampal volumes. Arch. Gen. Psychiatry 58(12), 1145–1151.
Magariños, A. M., and McEwen, B. S. (1995). Stress-induced atrophy of apical dendrites of hippocampal CA3c neurons: Comparison of stressors. Neuroscience. 69(1):83–88.
Mahaney, M. C., Williams-Blangero, S., Blangero, J., and Leland, M. M. (1993). Quantitative genetics
of relative organ weight variation in captive baboons. Hum. Biol 65(6), 991–1003.
Martorell, L., Valero, J., Mulet, B., Gutiérrez-Zotes, A., Cortés, M. J., Jariod, M., Pérez, M., Labad,
A., and Vilella, E. (2007). M129V variation in the prison protein gene and psychotic disorders:
Relationship to neuropsychological and psychopathological measures. J. Psychiatr. Res. 41(10),
885–892.
McDonald, C., and Murphy, K. C. (2003). The new genetics of schizophrenia. Psychiatry Clin. North Am.
26(1), 41–63. Review.
McGuffin, P., Famer, A. E., Gottesman, I. I., Murray, R. M., and Reveley, A. M. (1984). Twin
concordance for operationally defined schizophrenia. Confirmation of familiality and heritability.
Arch. Gen. Psychiatry 41(6), 541–545.
McIntosh, A. M., Baig, B. J., Hall, J., Job, D., Whalley, H. C., Lymer, G. K., Moorhead, T. W.,
Owens, D. G., Miller, P., Porteous, D., Lawrie, S. M., and Johnstone, E. C. (2007). Relationship of
catechol-O-methyltransferase variants to brain structure and function in a population at high risk
of psychosis. Biol. Psychiatry 61(10), 1127–1134.
McIntosh, A. M., Moorhead, T. W., Job, D., Lymer, G. K., Muñoz Maniega, S., McKirdy, J.,
Sussmann, J. E., Baig, B. J., Bastin, M. E., Porteous, D., Evans, K. L., Johnstone, E. C., et al.
(2008). The effects of a neuregulin 1 variant on white matter density and integrity. Mol. Psychiatry
13(11), 1054–1059.
McManus, I. C., and Bryden, M. P. (1991). Geschwind’s theory of cerebral lateralization: Developing a
formal, causal model. Psychol. Bull. 110(2), 237–253. Review.
McNeil, T. F. (2000). Perinatal risk factors and schizophrenia: Selective review and methodological
concerns. Epidemiol. Rev. 17(1), 107–112. Review.
Meisenzahl, E. M., Rujescu, D., Kirner, A., Giegling, I., Kathmann, N., Leinsinger, G., Maag, K.,
Hegerl, U., Hahn, K., and Möller, H. J. (2001). Association of an interleukin-1beta genetic polymorphism with altered brain structure in patients with schizophrenia. Am. J. Psychiatry 158(8), 1316–1319.
68
KAYMAZ AND VAN OS
Murray, R. M. (1994). Neurodevelopmental schizophrenia: The rediscovery of dementia praecox. Br.
J. Psychiatry 25(Suppl.), 6–12. Review.
Nadarajah, B., and Parnavelas, J. G. (2002). Modes of neuronal migration in the developing cerebral
cortex. Nat Rev Neurosci. 3(6), 423–432. Review.
Narr, K. L., van Erp, T. G., Cannon, T. D., Woods, R. P., Thompson, P. M., Jang, S., Blanton, R.,
Poutanen, V. P., Huttunen, M., Lönnqvist, J., Standerksjöld-Nordenstam, C. G., Kaprio, J., et al.
(2002). A twin study of genetic contributions to hippocampal morphology in schizophrenia.
Neurobiol. Dis. 11(1), 83–95.
Nguyen, G. H., Bouchard, J., Boselli MGTolstoi, L. G., Keith, L., Baldwin, C., Nguyen, N. C.,
Schultz, M., Herrera, V. L., and Smith, C. L. (2003). DNA stability and schizophrenia in twins.
Am. J. Med. Genet. B Neuropsychiatr. Genet. 120B(1), 1–10.
Ohnishi, T., Hashimoto, R., Mori, T., Nemoto, K., Moriguchi, Y., Iida, H., Noguchi, H.,
Nakabayashi, T., Hori, H., Ohmori, M., Tsukue, R., Anami, K., et al. (2006). The association
between the Val158Met polymorphism of the catechol-O-methyl transferase gene and morphological abnormalities of the brain in chronic schizophrenia. Brain 129(Pt 2), 399–410.
Okugawa, G., Nobuhara, K., Takase, K., and Kinoshita, T. (2007). Cerebellar posterior superior
vermis and cognitive cluster scores in drug-naive patients with first-episode schizophrenia.
Neuropsychobiology 56(4), 216–219.
Oppenheim, J. S., Skerry, J. E., Tramo, M. J., and Gazzaniga, M. S. (1989). Magnetic resonance
imaging morphology of the corpus callosum in monozygotic twins. Ann. Neurol. 26(1), 100–104.
Papiol, S., Molina, V., Desco, M., Rosa, A., Reig, S., Gispert, J. D., Sanz, J., Palomo, T., and
Fañanás, L. (2005). Ventricular enlargement in schizophrenia is associated with a genetic
polymorphism at the interleukin-1 receptor antagonist gene. Neuroimage 27(4), 1002–1006.
Pearlson, G. D., and Calhoun, V. (2007). Structural and functional magnetic resonance imaging in
psychiatric disorders. Can J. Psychiatry 52(3), 158–166. Review.
Pennington, B. F., Filipek, P. A., Lefly, D., Chhabildas, N., Kennedy, D. N., Simon, J. H., Filley, C. M.,
Galaburda, A., and DeFries, J. C. (2000). A twin MRI study of size variations in human brain.
J. Cogn. Neurosci. 12(1), 223–232.
Peper, J. S., Schnack, H. G., Brouwer, R. M., Van Baal, G. C., Pjetri, E., Székely, E., van
Leeuwen, M., van den Berg, S. M., Collins, D. L., Evans, A. C., Boomsma, D. I., Kahn, R. S.,
et al. (2009). Heritability of regional and global brain structure at the onset of puberty: A magnetic
resonance imaging study in 9-year-old twin pairs. Hum. Brain Mapp. 30(7), 2184–2196.
Petronis, A., and Kennedy, J. L. (1995). Unstable genes: Unstable mind? Am. J. Psychiatry 152(2), 164–
172. Review.
Petronis, A., Gottesman, I. I., Kan, P., Kennedy, J. L., Basile, V. S., Paterson, A. D., and
Popendikyte, V. (2003). Monozygotic twins exhibit numerous epigenetic differences: Clues to
twin discordance? Schizophr. Bull. 29(1), 169–178.
Pezawas, L., Verchinski, B. A., Mattay, V. S., Callicott, J. H., Kolachana, B. S., Straub, R. E.,
Egan, M. F., Meyer-Lindenberg, A., and Weinberger, D. R. (2004). The brain-derived neurotrophic factor val66met polymorphism and variation in human cortical morphology. J. Neurosci.
24(45), 10099–10102.
Pfefferbaum, A., Sullivan, E. V., and Carmelli, D. (2004). Morphological changes in aging brain
structures are differentially affected by time-linked environmental influences despite strong
genetic stability. Neurobiol. Aging 25(2), 175–183.
Plassman, B. L., Welsh-Bohmer, K. A., Bigler, E. D., Johnson, S. C., Anderson, C. V., Helms, M. J.,
Saunders, A. M., and Breitner, J. C. (1997). Apolipoprotein E epsilon 4 allele and hippocampal
volume in twins with normal cognition. Neurology 48(4), 985–989.
Posthuma, D., and Boomsma, D. I. (2000). A note on the statistical power in extended twin designs.
Behav. Genet. 30(2), 147–158.
69
HERITABILITY OF STRUCTURAL BRAIN TRAITS
Posthuma, D., de Geus, E. J., Neale, M. C., Hulshoff Pol, H. E., Baaré, W. F., Kahn, R. S., and
Boomsma, D. (2000). Multivariate genetic analysis of brain structure in an extended twin design.
Behav. Genet. 30(4), 311–319.
Posthuma, D., Mulder, E. J., Boomsma, D. I., and de Geus, E. J. (2002a). Genetic analysis of IQ,
processing speed and stimulus-response incongruency effects. Biol. Psychol 61(1–2), 157–182.
Posthuma, D., De Geus, E. J., Baaré, W. F., Hulshoff Pol, H. E., Kahn, R. S., and Boomsma, D. I.
(2002b). The association between brain volume and intelligence is of genetic origin. Nat. Neurosci.
5(2), 83–84.
Posthuma, D., Baaré, W. F., Hulshoff Pol, H. E., Kahn, R. S., Boomsma, D. I., and De Geus, E. J. (2003).
Genetic correlations between brain volumes and the WAIS-III dimensions of verbal comprehension, working memory, perceptual organization, and processing speed. Twin Res. 6(2), 131–139.
Prasad, S., Chowdari, K. V., Wood, J., Bhatia, T., Deshpande, S. N., Nimgaonkar, V. L., and
Thelma, B. K. (2005). Association analysis of NOTCH 4 polymorphisms with schizophrenia
among two independent family based samples. Am. J. Med. Genet. B Neuropsychiatr. Genet. 131B(1),
6–9.
Price, B. (1950). Primary biases in twin studies; a review of prenatal and natal difference-producing
factors in monozygotic pairs. Am. J. Hum. Genet. 2(4), 293–352.
Porteous, D. (2008). Genetic causality in schizophrenia and bipolar disorder: Out with the old and in
with the new. Curr. Opin Genet. Dev. 18(3), 229–234. Review.
Rais, M., Cahn, W., Van Haren, N., Schnack, H., Caspers, E., Hulshoff Pol, H., and Kahn, R. (2008).
Excessive brain volume loss over time in cannabis-using first-episode schizophrenia patients. Am.
J. Psychiatry 165(4), 490–496.
Rakic, P. (1972). Mode of cell migration to the superficial layers of fetal monkey neocortex. J. Comp.
Neurol. 145(1), 61–83.
Rakic, P. (1988). Specification of cerebral cortical areas. Science 241(4862), 170–176. Review.
Reveley, A. M., Reveley, M. A., Chitkara, B., et al. (1984). The genetic basis of cerebral ventricular
volume. Psychiatry Res. 13(3), 261–266.
Rijsdijk, F. V., van Haren, N. E., Picchioni, M. M., McDonald, C., Toulopoulou, T., Hulshoff
Pol, H. E., Kahn, R. S., Murray, R., and Sham, P. C. (2005). Brain MRI abnormalities in
schizophrenia: Same genes or same environment? Psychol. Med. 35(10), 1399–1409.
Rogers, J., Kochunov, P., Lancaster, J., Shelledy, W., Glahn, D., Blangero, J., and Fox, P. (2007).
Heritability of brain volume, surface area and shape: An MRI study in an extended pedigree of
baboons. Hum. Brain Mapp. 28(6), 576–583.
Ross, C. A., Margolis, R. L., Reading, S. A., Pletnikov, M., and Coyle, J. T. (2006). Neurobiology of
schizophrenia. Neuron 52(1), 139–153. Review.
Roth, G., and Dicke, U. (2005). Evolution of the brain and intelligence. Trends Cogn. Sci. 9(5), 250–257.
Rubenstein, J. L., and Rakic, P. (1999). Genetic control of cortical development. Cereb. Cortex 9(6),
521–523.
Rubenstein, J. L., Anderson, S., Shi, L., Miyashita-Lin, E., Bulfone, A., and Hevner, R. (1999).
Genetic control of cortical regionalization and connectivity. Cereb. Cortex 9(6), 524–532.
Rujescu, D., Meisenzahl, E. M., Giegling, I., Kirner, A., Leinsinger, G., Hegerl, U., Hahn, K., and
Möller, H. J. (2002). Methionine homozygosity at codon 129 in the prion protein is associated
with white matter reduction and enlargement of CSF compartments in healthy volunteers and
schizophrenic patients. Neuroimage 15(1), 200–206.
Rujescu, D., Meisenzahl, E. M., Krejcova, S., Giegling, I., Zetzsche, T., Reiser, M., Born, C. M.,
Möller, H. J., Veske, A., Gal, A., and Finckh, U. (2007). Plexin B3 is genetically associated with
verbal performance and white matter volume in human brain. Mol. Psychiatry 12(2), 190–194. 115.
Sapolsky, R. M., Uno, H., Rebert, C. S., and Finch, C. E. (1990). Hippocampal damage associated
with prolonged glucocorticoid exposure in primates. J. Neurosci. 10(9), 2897–2902.
70
KAYMAZ AND VAN OS
Sánchez, M. M., Hearn, E. F., Do, D., Rilling, J. K., and Herndon, J. G. (1998). Differential rearing
affects corpus callosum size and cognitive function of rhesus monkeys. Brain Res 812(1–2), 38–49.
Sapolsky, R. M. (2000). Glucocorticoids and hippocampal atrophy in neuropsychiatric disorders. Arch.
Gen. Psychiatry 57(10), 925–935.
Scamvougeras, A., Kigar, D. L., Jones, D., Weinberger, D. R., and Witelson, S. F. (2003). Size of the
human corpus callosum is genetically determined: An MRI study in mono and dizygotic twins.
Neurosci. Lett. 338(2), 91–94.
Schmitt, J. E., Lenroot, R. K., Wallace, G. L., Ordaz, S., Taylor, K. N., Kabani, N., Greenstein, D.,
Lerch, J. P., Kendler, K. S., Neale, M. C., and Giedd, J. N. (2008). Identification of genetically
mediated cortical networks: A multivariate study of pediatric twins and siblings. Cereb. Cortex 18(8),
1737–1747.
Semendeferi, K., Damasio, H., Frank, R., and Van Hoesen, G. W. (1997). The evolution of the frontal
lobes: A volumetric analysis based on three-dimensional reconstructions of magnetic resonance
scans of human and ape brains. J. Hum. Evol. 32(4), 375–388.
Semendeferi, K., Lu, A., Schenker, N., Damasio, H., and Damasio, H. (2001). Humans and great apes
share a large frontal cortex. Nat. Neurosci. 5(3), 272–276.
Siegel, S. J., Ginsberg, S. D., Hof, P. R., Foote, S. L., Young, W. G., Kraemer, G. W.,
McKinney, W. T., and Morrison, J. H. (1993). Effects of social deprivation in prepubescent rhesus
monkeys: Immunohistochemical analysis of the neurofilament protein triplet in the hippocampal
formation. Brain Res. 619(1–2), 299–305.
Silventoinen, K. (2003). Determinants of variation in adult body height. J. Biosoc. Sci. 35(2), 263–285.
Singh, S. M., Murphy, B., and O’Reilly, R. (2002). Epigenetic contributors to the discordance of
monozygotic twins. Clin. Genet. 62(2), 97–103. Review.
Sheline, Y. I., Sanghavi, M., Mintun, M. A., and Gado, M. H. (1999). Depression duration but not age
predicts hippocampal volume loss in medically healthy women with recurrent major depression.
J. Neurosci. 19(12), 5034–5043.
Soininen, H. S., and Riekkinen, P. J. Sr. (1996). Apolipoprotein E, memory and Alzheimer’s disease.
Trends Neurosci. 19(6), 224–228.
Spence, M. A., Westlake, J., and Lange, K. (1977). Estimation of the variance components for dermal
ridge count. Ann. Hum. Genet. 41(1), 111–115.
Steinmetz, H., Herzog, A., Schlaug G Huang, Y., and Jäncke, L. (1995). Brain (A) symmetry in
monozygotic twins. Cereb. Cortex 5(4), 296–300.
Suddath, R. L., Christison, G. W., Torrey, E. F., Casanova, M. F., and Weinberger, D. R. (1990).
Anatomical abnormalities in the brains of monozygotic twins discordant for schizophrenia.
N. Engl. J. Med. 322(12), 789–794.
Sullivan, E. V., Pfefferbaum, A., Swan, G. E., and Carmelli, D. (2001). Heritability of hippocampal size
in elderly twin men: Equivalent influence from genes and environment. Hippocampus 11(6),
754–762.
Sullivan, P. F., Kendler, K. S., and Neale, M. C. (2003). Schizophrenia as a complex trait: Evidence
from a meta-analysis of twin studies. Arch. Gen. Psychiatry 60(12), 1187–1192.
Szeszko, P. R., Lipsky, R., Mentschel, C., Robinson, D., Gunduz-Bruce, H., Sevy, S., Ashtari, M.,
Napolitano, B., Bilder, R. M., Kane, J. M., Goldman, D., and Malhotra, A. K. (2005). Brainderived neurotrophic factor val66met polymorphism and volume of the hippocampal formation.
Mol. Psychiatry 10(7), 631–636.
Tohgi, H., Takahashi, S., Kato, E., Homma, A., Niina, R., Sasaki, K., Yonezawa, H., and Sasaki, M.
(1997). Reduced size of right hippocampus in 39- to 80-year-old normal subjects carrying the
apolipoprotein E epsilon4 allele. Neurosci. Lett. 236(1), 21–24.
Thompson, P. M., and Toga, A. W. (1997). Detection, visualization and animation of abnormal
anatomic structure with a deformable probabilistic brain atlas based on random vector field
transformations. Med. Image Anal. 1(4), 271–294.
71
HERITABILITY OF STRUCTURAL BRAIN TRAITS
Thompson, P. M., Schwartz, C., and Toga, A. W. (1996a). High-resolution random mesh algorithms
for creating a probabilistic 3D surface atlas of the human brain. Neuroimage 3(1), 19–34.
Thompson, P. M., Schwartz, C., Lin, R. T., Khan, A. A., and Toga, A. W. (1996b). Three-dimensional
statistical analysis of sulcal variability in the human brain. J. Neurosci. 16(13), 4261–4274.
Thompson, P. M., Cannon, T. D., Narr, K., van Erp, T., Poutanen, V. P., Huttunen, M., Lönnqvist, J.,
Standertskjöld-Nordenstam, C. G., Kaprio, J., Khaledy, M., Dail, R., Zoumalan, C. I., et al.
(2001). Genetic influences on brain structure. Nat. Neurosci. 4(12), 1253–1258.
Thompson, P., Cannon, T. D., and Toga, A. W. (2002). Mapping genetic influences on human brain
structure. Ann. Med. 34(7–8), 523–536. Review.
Tramo, M. J., Loftus, W. C., Stukel, T. A., Green, R. L., Weaver, J. B., and Gazzaniga, M. S. (1998).
Brain size, head size, and intelligence quotient in monozygotic twins. Neurology 50(5), 1246–1252.
Tsujita, T., Niikawa, N., Yamashita, H., Imamura, A., Hamada, A., Nakane, Y., and Okazaki, Y.
(1998). Genomic discordance between monozygotic twins discordant for schizophrenia. Am.
J. Psychiatry 155(3), 422–424.
Uno, H., Tarara, R., Else, J. G., Suleman, M. A., and Sapolsky, R. M. (1989). Hippocampal damage
associated with prolonged and fatal stress in primates. J. Neurosci. 9(5), 1705–1711.
van Erp, T. G., Saleh, P. A., Huttunen, M., Lönnqvist, J., Kaprio, J., Salonen, O., Valanne, L.,
Poutanen, V. P., Standertskjöld-Nordenstam, C. G., and Cannon, T. D. (2004). Hippocampal
volumes in schizophrenic twins. Arch. Gen. Psychiatry 61(4), 346–353.
van Haren, N. E., Picchioni, M. M., McDonald, C., Marshall, N., Davis, N., Ribchester, T., Hulshoff
Pol, H. E., Sharma, T., Sham, P., Kahn, R. S., and Murray, R. (2004). A controlled study of brain
structure in monozygotic twins concordant and discordant for schizophrenia. Biol. Psychiatry 56(6),
454–461.
van Haren, N. E., Hulshoff Pol, H. E., Schnack, H. G., Cahn, W., Mandl, R. C., Collins, D. L.,
Evans, A. C., and Kahn, R. S. (2007). Focal gray matter changes in schizophrenia across the
course of the illness: A 5-year follow-up study. Neuropsychopharmacology 32(10), 2057–2066.
van Haren, N. E., Bakker, S. C., and Kahn, R. S. (2008). Genes and structural brain imaging in
schizophrenia. Curr. Opin. Psychiatry 21(2), 161–167. Review.
van Os, J. (2009). A salience dysregulation syndrome. Br. J. Psychiatry 194(2), 101–103.
van Os, J., and Murray, R. (2008). Gene-environment interactions in schizophrenia. Introduction.
Schizophr. Bull. 34(6), 1064–1065.
van Os, J., Rutten, B. P., and Poulton, R. (2008). Gene-environment interactions in schizophrenia:
Review of epidemiological findings and future directions. Schizophr. Bull. 34(6), 1066–1082.
Verdoux, H., Geddes, J. R., Takei, N., Lawrie, S. M., Bovet, P., Eagles, J. M., Heun, R.,
McCreadie, R. G., McNeil, T. F., O’Callaghan, E., Stöber, G., Willinger, M. U., et al. (1997).
Obstetric complications and age at onset in schizophrenia: An international collaborative metaanalysis of individual patient data. Am. J. Psychiatry 154(9), 1220–1227.
Vincent, J. B., Kalsi, G., Klempan, T., Tatuch, Y., Sherrington, R. P., Breschel, T., McInnis, M. G.,
Brynjolfsson, J., Petursson, H., Gurling, H. M., Gottesman, I. I., Torrey, E. F., et al. (1998). No
evidence of expansion of CAG or GAA repeats in schizophrenia families and monozygotic twins.
Hum. Genet. 103(1), 41–47.
Wallace, G. L., Eric Schmitt, J., Lenroot, R., Viding, E., Ordaz, S., Rosenthal, M. A., Molloy, E. A.,
Clasen, L. S., Kendler, K. S., Neale, M. C., and Giedd, J. N. (2006). A pediatric twin study of
brain morphometry. J. Child Psychol. Psychiatry 47(10), 987–993.
Wang, J. K., Li, Y., and Su, B. (2008). A common SNP of MCPH1 is associated with cranial volume
variation in Chinese population. Hum. Mol. Genet. 17(9), 1329–1335.
Ward, K. E., Friedman, L., Wise, A., and Schulz, S. C. (1996). Meta-analysis of brain and cranial size
in schizophrenia. Schizophr. Res. 22(3), 197–213.
Wassink, T. H., Nelson, J. J., Crowe, R. R., and Andreasen, N. C. (1999). Heritability of BDNF alleles
and their effect on brain morphology in schizophrenia. Am. J. Med. Genet. 88(6), 724–728.
72
KAYMAZ AND VAN OS
Wassink, T. H., Crowe, R. R., and Andreasen, N. C. (2000). Tumor necrosis factor receptor-II:
Heritability and effect on brain morphology in schizophrenia. Mol. Psychiatry 5(6), 678–682.
Wassink, T. H., Nopoulos, P., Pietila, J., Crowe, R. R., and Andreasen, N. C. (2003). NOTCH4 and
the frontal lobe in schizophrenia. Am. J. Med. Genet. B Neuropsychiatr. Genet. 118B(1), 1–7.
Weinberger, D. R. (1995). From neuropathology to neurodevelopment. Lancet 346(8974), 552–557.
Review.
White, T., Andreasen, N. C., and Nopoulos, P. (2002). Brain volumes and surface morphology in
monozygotic twins. Cereb Cortex 12(5), 486–493.
Williams, R. R., Hunt, S. C., Hasstedt, S. J., Hopkins, P. N., Wu, L. L., Berry, T. D., Stults, B. M.,
Barlow, G. K., and Kuida, H. (1990). Genetics of hypertension: What we know and don’t know.
Clin. Exp. Hypertens. A. 12(5), 865–876.
Williams, H. J., Owen, M. J., and O’Donovan, M. C. (2007). Is COMT a susceptibility gene for
schizophrenia? Schizophr. Bull. 33(3), 635–641.
Woods, R. P., Freimer, N. B., De Young, J. A., Fears, S. C., Sicotte, N. L., Service, S. K.,
Valentino, D. J., Toga, A. W., and Mazziotta, J. C. (2006). Normal variants of Microcephalin
and ASPM do not account for brain size variability. Hum. Mol. Genet. 15(12), 2025–2029.
Wright, I. C., Rabe-Hesketh, S., and Woodruff, P. W. (2000). Meta-analysis of regional brain volumes
in schizophrenia. Am. J. Psychiatry 157(1), 16–25.
Wright, I. C., Sham, P., Murray, R. M., Weinberger, D. R., and Bullmore, E. T. (2002). Genetic
contributions to regional variability in human brain structure: Methods and preliminary results.
Neuroimage 17(1), 256–271.
Zinkstok, J., Schmitz, N., van Amelsvoort, T., de Haan, L., Yusuf, M. A., Baas, F., and Linszen, D.
(2007). Genetic variation in COMT and PRODH is associated with brain anatomy in patients
with schizophrenia. Genes Brain Behav. 7(1), 61–69.
73
Acta Psychiatr Scand 2009: 120: 249–252
All rights reserved
DOI: 10.1111/j.1600-0447.2009.01472.x
! 2009 John Wiley & Sons A/S
ACTA PSYCHIATRICA
SCANDINAVICA
Editorial
Murray et al. (2004) revisited: is bipolar
disorder identical to schizophrenia without
developmental impairment?
In 2004, Murray et al. (1) suggested that the main
difference between bipolar disorder (BPD) and
schizophrenia (Sz) was a larger prevalence of
developmental impairments in the latter. A
number of recent articles in Acta Psychiatrica
Scandinavica suggest that categories of affective
and non-affective psychosis may be similar in their
pattern of associations with measures as diverse as
social cognition (2), brain volumes (3) and metabolic dysregulation (4), begging the question what
the actual status is of the prediction offered by
Murray et al. more than 5 years ago. These issues
are particularly important given the upcoming
revisions of diagnostic systems in psychiatry: how
should bipolar disorder and schizophrenia be
classified in DSM-V and ICD-11?
A review was conducted of studies focussing on
this issue, with a special focus on work published in
the last 10 years; results are displayed in Table 1
below.
Both Sz and BPD have onset in adolescence and
young adulthood, with an earlier onset in men. The
incidence of both disorders is low and associated
with a high prevalence:incidence ratio, indicative of
a high rate of chronicity. Schizophrenia patients
Fig. 1. A combined dimensional and
categorical system of diagnosis of psychotic disorders. Categorical diagnoses
of schizophrenia (yellow), bipolar disorder (brown), and schizoaffective disorder (pink) are accompanied by a
patient!s quantitative scores (connected
by red lines) on five main dimensions of
psychopathology.
74
have high rates of affective symptoms, although
lower than in BPD, and bipolar patients have
negative symptoms and cognitive impairment,
although lower than in Sz. The diagnostic contrast
between Sz and BPD is high for affective symptoms, but this is mostly due to artificial diagnostic
exclusion criteria.
Sz and BPD have correlated genetic liabilities,
some of which is becoming substantiated in recent
genome-wide molecular genetic association studies.
Of interest, however, is the fact that genetic risk for
Sz is strongly expressed as neurocognitive impairment whereas genetic risk for BPD is only weakly
expressed in the neurocognitive domain. Danish
studies have shown that BPD and Sz have a high
population comorbidity index; general population
studies have shown that both BPD and Sz phenotypes are associated with psychometric risk states
in healthy individuals, and that BPD and Sz
psychometric risk states are similarly highly comorbid with each other. Risk factors representing
social stress and defeat as well as early emotional
alterations are associated with both BPD and Sz
whereas risk factors reflecting early motor and
cognitive alterations appear specific for schizo-
Neurocognition
Mentalizing ability
Neurocognitive impairment
Genetics
Cognition
Intermediary phenotypes
Neuroradiological findings and
other biological markers
Developmental alterations
Risk factors
Epidemiology
Schizophrenia (Sz)
Generalised alterations associated
with familial ⁄ genetic risk for Sz(40, 41)
Impaired(37, 38)
Impaired(2)
Impaired(37, 38)
Impaired(2)
Molecular genetic studies (GWAS) suggest
genetic variation common to Sz and BPD
e.g. ZNF804A and CACNA1C(36)
Sz>BPD
Sz>BPD; may be mediated by neurocognitive
alterations(2, 39)
Other measures, e.g. neurophysiology, too
few comparative data, some reports
suggest similarities(43)
Generally much fewer studies in BPD;
meta-analyses largely in Sz only.
Sz>BPD, diagnostic contrast low(12)
Sz>BPD, diagnostic contrast low(12)
Sz>BPD, diagnostic contrast low(12)
BPD>Sz, diagnostic contrast high(12, 13)
BPD>Sz, diagnostic contrast high(12, 13)
High P:I ratio indicates high proportion chronic
course; Onset and sex differences shifted to
later ages in BPD compared to Sz
Comment
Strong comorbidity between psychosis and
bipolar psychometric risk states in
population (8), similar to the comorbidity
seen in clinical disorders (9)
Bipolar disorder (BPD)
Subclinical bipolarity prevalence around
10–20% (7)
At a rate of around 3% over 1 year (11)
Present
Present
Present
Present
Present
Present
Present
Present
Present
Present
!1%
!1%
!100
!100
Developmental pattern(14)
Developmental pattern(15)
Earlier in men(16)
Earlier in men(17)
Positive association(18)
Positive association(19)
Positive association(20)
No association(21, 22)
Positive association(23)
Positive association(24, 25)
Positive association(26)
Positive association(27, 28)
Positive association (early)(29)
Positive association (late) (30)
Reviews suggest association with Sz(31) but not BPD(32). However, reviews largely based on
retrospective studies; suggestion of increased risk for BPD in rare prospective studies(30, 33, 34)
Positive association(35)
Positive association(35)
Positive association(35)
Positive association(35)
Subclinical psychosis prevalence around
5–10% (6)
At a rate of around 4% over 1 year
follow-up (10)
Very few and specific alterations
associated with familial risk for
BPD(37, 42)
Child cognitive ability
Alterations (44–46)
No alterations(47–49)
Child motor development
Alterations(44–46)
No alterations(47–49)
Child emotional development
Alterations(44–46)
Alterations(47–49)
Many individual studies but to date no systematic quantitative integration of findings of Sz vs. BPD. Uncertain whether this will ever be possible given heterogeneity of measures, methodologies,
patient selection, unmeasured confounding and differential neurotrophic effects of medications(50). Fronto-temporal systems may be most affected in both Sz and BPD. Ventricular enlargement,
cortical gray matter reductions and abnormal gyral folding (51) reported in both Sz and BPD, data suggest Sz>BPD; suggestion of qualitative differences in hippocampal and amygdalar volumes in Sz
and BPD.
Relative with BPD
Relative with Sz
Psychometric risk states in general
population
Psychopathology
Variable
Subclinical psychometric risk states in
representative population samples
Transition from subclinical psychometric
risk states to clinical disorder in
epidemiological studies
Positive symptoms
Negative symptoms
Disorganisation
Mania symptoms
Depression symptoms
Lifetime prevalence
Prevalence:incidence ratio
Age of onset
Sex differences onset
Cannabis
Urban upbringing
Ethnic group
Parental age
Prenatal famine
Pregnancy complications
Type of variable
Table 1. Review of critical contrasts and similarities in schizophrenia and bipolar disorder
Editorial
75
Editorial
phrenia. Interesting is the differential association
with growing up in an urban environment (not
associated with bipolar disorder), suggesting that
this exposure impacts on specific developmental
alterations associated with schizophrenia.
The literature on neuroimaging, biological variables and prenatal life suggests differences and
similarities as regards their prevalences in Sz and
BPD; firm conclusions, however, cannot be drawn
because of many methodological constraints and
the lack of direct comparisons.
In conclusion, science is catching up with century-old diagnostic traditions in psychiatry. It is
becoming apparent that due to the strict separation
imposed by DSM and ICD on the domains of
affective and non-affective psychotic disorder,
major opportunities have been missed to study
the causes and treatment of psychiatric disorders.
Psychotic disorders appear to be originating from
(partly) overlapping areas of risk, one more developmental associated with cognitive impairment,
and one more associated with affective dysregulation. It has been suggested that where these areas
of risk overlap and interact with environmental
risks, delusions and hallucinations may ensue
through a final common pathway of dopamine
dysregulation and affective cognitive biases (5).
Therefore an elegant way of adapting the revisions
of DSM and ICD diagnostic manuals to the
scientific reality is to keep diagnostic categories
more or less as they were and add cross-disorder
dimensions of psychopathology, as depicted in
Fig. 1. The simple addition of dimensions would
finally clear the way for an official cross-diagnostic
approach in psychiatry, ending a dysfunctional and
rigid system of artificial partitioning.
Acta Psychiatrica Scandinavica
Nil Kaymaz and Jim van Os
Invited Guest Editors
Acknowledgement
Nil Kaymaz is supported by the Netherlands Organisation
for Scientific Research (NWO) under project number:
017.002.048.
References
1. Murray RM, Sham P, Van Os J, Zanelli J, Cannon M,
McDonald C. A developmental model for similarities and
dissimilarities between schizophrenia and bipolar disorder.
Schizophr Res 2004;71:405–416.
2. Bora E, Yucel M, Pantelis C. Theory of mind impairment:
a distinct trait-marker for schizophrenia spectrum disorders and bipolar disorder? Acta Psychiatr Scand
2009;120:253–264.
76
3. Sun J, Maller JJ, Daskalakis ZJ, Furtado CC, Fitzgerald
PB. Morphology of the corpus callosum in treatmentresistant schizophrenia and major depression. Acta Psychiatr Scand 2009;120:265–273.
4. Huang MC, Lu ML, Tsai CJ et al. Prevalence of metabolic
syndrome among patients with schizophrenia or schizoaffective disorder in Taiwan. Acta Psychiatr Scand
2009;120:274–280.
5. Van Os J, Kapur S. Schizophrenia. Lancet 2009;374:635–645.
6. Van Os J, Linscott RJ, Myin-Germeys I, Delespaul P,
Krabbendam L. A systematic review and meta-analysis of
the psychosis continuum: evidence for a psychosis proneness-persistence-impairment model of psychotic disorder.
Psychol Med 2009;39:179–195.
7. Tijssen MJA, Van Os J, Wittchen HU et al. Evidence that
bipolar disorder is the poor outcome fraction of a common
developmental phenotype: an 8-year cohort study in young
people. Psychol Med 2009;11:1–11.
8. van Rossum I, Dominguez M, Lieb R, Wittchen HU, van Os
J. Affective dysregulation and reality distortion: a 10-year
prospective study of their association and clinical relevance. Submitted manuscript 2009.
9. Laursen TM, Agerbo E, Pedersen CB. Bipolar disorder,
schizoaffective disorder, and schizophrenia overlap: a new
comorbidity index. J Clin Psychiatry 2009;[Epub ahead of
print].
10. Hanssen M, Bak M, Bijl R, Vollebergh W, van Os J. The
incidence and outcome of subclinical psychotic experiences
in the general population. Br J Clin Psychol 2005;44:181–
191.
11. Regeer EJ, Krabbendam L, de Graaf R, ten Have M, Nolen
WA, van Os J. A prospective study of the transition rates of
subthreshold (hypo)mania and depression in the general
population. Psychol Med 2006;36:619–627.
12. Van Os J, Gilvarry C, Bale R et al. Diagnostic value of the
DSM and ICD categories of psychosis: An evidence-based
approach. Soc Psychiatry Psychiatr Epidemiol 2000;35:
305–311.
13. Peralta V, Cuesta MJ. Characterization of affective domains within the nonaffective psychotic disorders. Schizophr Res 2009;111:61–69.
14. McGrath J, Saha S, Chant D, Welham J. Schizophrenia: a
concise overview of incidence, prevalence, and mortality.
Epidemiol Rev 2008;30:67–76.
15. Kennedy N, Everitt B, Boydell J, Van Os J, Jones PB,
Murray RM. Incidence and distribution of first-episode
mania by age: results from a 35-year study. Psychol Med
2005;35:855–863.
16. Castle DJ, Murray RM. The neurodevelopmental basis of
sex differences in schizophrenia. Psychol Med 1991;21:
565–575.
17. Kennedy N, Boydell J, Kalidindi S et al. Gender differences
in incidence and age at onset of mania and bipolar disorder
over a 35-year period in Camberwell, England. Am J
Psychiatry 2005;162:257–262.
18. Henquet C, Murray R, Linszen D, van Os J. The environment and schizophrenia: the role of cannabis use. Schizophr Bull 2005;31:608–612.
19. Henquet C, Krabbendam L, de Graaf R, ten Have M, van Os
J. Cannabis use and expression of mania in the general
population. J Affect Disord 2006;95:103–110.
20. Krabbendam L, van Os J. Schizophrenia and urbanicity: a
major environmental influence–conditional on genetic risk.
Schizophr Bull 2005;31:795–799.
21. Mortensen PB, Pedersen CB, Melbye M, Mors O, Ewald H.
Individual and familial risk factors for bipolar affective
Editorial
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
disorders in Denmark. Arch Gen Psychiatry 2003;60:1209–
1215.
Kaymaz N, Krabbendam L, de Graaf R, Nolen W, Ten Have
M, van Os J. Evidence that the urban environment specifically impacts on the psychotic but not the affective
dimension of bipolar disorder. Soc Psychiatry Psychiatr
Epidemiol 2006;41:679–685.
Cantor-Graae E, Selten JP. Schizophrenia and migration: a
meta-analysis and review. Am J Psychiatry 2005;162:12–24.
Van Os J, Castle DJ, Takei N, Der G, Murray RM. Psychotic illness in ethnic minorities: clarification from the
1991 census. Psychol Med 1996;26:203–208.
Kirkbride JB, Fearon P, Morgan C et al. Heterogeneity in
incidence rates of schizophrenia and other psychotic syndromes: findings from the 3-center AeSOP study. Arch
Gen Psychiatry 2006;63:250–258.
Wohl M, Gorwood P. Paternal ages below or above
35 years old are associated with a different risk of schizophrenia in the offspring. Eur Psychiatry 2007;22:22–26.
Menezes PR, Lewis G, Rasmussen F et al. Paternal and
maternal ages at conception and risk of bipolar affective
disorder in their offspring. Psychol Med 2009;[Epub ahead
of print].
Frans EM, Sandin S, Reichenberg A, Lichtenstein P, Langstrom N, Hultman CM. Advancing paternal age and
bipolar disorder. Arch Gen Psychiatry 2008;65:1034–1040.
Susser E, Neugebauer R, Hoek HW et al. Schizophrenia
after prenatal famine. Further evidence [see comments].
Arch Gen Psychiatry 1996;53:25–31.
Brown AS, Van Os J, Driessens C, Hoek HW, Susser ES.
Further evidence of relation between prenatal famine and
major affective disorder. Am J Psychiatry 2000;157:190–
195.
Cannon M, Jones PB, Murray RM. Obstetric complications
and schizophrenia: historical and meta-analytic review.
Am J Psychiatry 2002;159:1080–1092.
Scott J, McNeill Y, Cavanagh J, Cannon M, Murray R.
Exposure to obstetric complications and subsequent
development of bipolar disorder: systematic review. Br J
Psychiatry 2006;189:3–11.
Zornberg GL, Buka SL, Tsuang MT. Hypoxic-ischemiarelated fetal ⁄ neonatal complications and risk of schizophrenia and other nonaffective psychoses: a 19-year
longitudinal study. Am J Psychiatry 2000;157:196–202.
Laursen TM, Munk-Olsen T, Nordentoft M, Bo Mortensen
P. A comparison of selected risk factors for unipolar
depressive disorder, bipolar affective disorder, schizoaffective disorder, and schizophrenia from a Danish population-based cohort. J Clin Psychiatry 2007;68:1673–1681.
Van Snellenberg JX, de Candia T. Meta-analytic evidence
for familial coaggregation of schizophrenia and bipolar
disorder. Arch Gen Psychiatry 2009;66:748–755.
O!Donovan MC, Craddock NJ, Owen MJ. Genetics of
psychosis; insights from views across the genome. Hum
Genet 2009;126:3–12.
37. Arts B, Jabben N, Krabbendam L, van Os J. Meta-analyses
of cognitive functioning in euthymic bipolar patients and
their first-degree relatives. Psychol Med 2008;38:771–785.
38. Krabbendam L, Arts B, van Os J, Aleman A. Cognitive
functioning in patients with schizophrenia and bipolar
disorder: a quantitative review. Schizophr Res
2005;80:137–149.
39. van Hooren S, Versmissen D, Janssen I et al. Social cognition and neurocognition as independent domains in psychosis. Schizophr Res 2008;103:257–265.
40. Toulopoulou T, Picchioni M, Rijsdijk F et al. Substantial
genetic overlap between neurocognition and schizophrenia:
genetic modeling in twin samples. Arch Gen Psychiatry
2007;64:1348–1355.
41. Dickinson D, Ramsey ME, Gold JM. Overlooking the
obvious: a meta-analytic comparison of digit symbol coding tasks and other cognitive measures in schizophrenia.
Arch Gen Psychiatry 2007;64:532–542.
42. Bora E, Yucel M, Pantelis C. Cognitive endophenotypes of
bipolar disorder: a meta-analysis of neuropsychological
deficits in euthymic patients and their first-degree relatives.
J Affect Disord 2009;113:1–20.
43. Sanchez-Morla EM, Garcia-Jimenez MA, Barabash A et al.
P50 sensory gating deficit is a common marker of vulnerability to bipolar disorder and schizophrenia. Acta Psychiatr Scand 2008;117:313–318.
44. Welham J, Isohanni M, Jones P, McGrath J. The antecedents of schizophrenia: a review of birth cohort studies.
Schizophr Bull 2009;35:603–623.
45. Woodberry KA, Giuliano AJ, Seidman LJ. Premorbid IQ in
schizophrenia: a meta-analytic review. Am J Psychiatry
2008;165:579–587.
46. Tarbox SI, Pogue-Geile MF. Development of social functioning in preschizophrenia children and adolescents: a
systematic review. Psychol Bull 2008;134:561–583.
47. Cannon M, Caspi A, Moffitt TE et al. Evidence for earlychildhood, pan-developmental impairment specific to
schizophreniform disorder: results from a longitudinal
birth cohort. Arch Gen Psychiatry 2002;59:449–456.
48. Reichenberg A, Weiser M, Rabinowitz J et al. A populationbased cohort study of premorbid intellectual, language,
and behavioral functioning in patients with schizophrenia,
schizoaffective disorder, and nonpsychotic bipolar disorder. Am J Psychiatry 2002;159:2027–2035.
49. Zammit S, Allebeck P, David AS et al. A longitudinal study of
premorbid IQ Score and risk of developing schizophrenia,
bipolar disorder, severe depression, and other nonaffective
psychoses. Arch Gen Psychiatry 2004;61:354–360.
50. Gur RE, Keshavan MS, Lawrie SM. Deconstructing psychosis with human brain imaging. Schizophr Bull
2007;33:921–931.
51. McIntosh AM, Moorhead TW, McKirdy J et al. Prefrontal
gyral folding and its cognitive correlates in bipolar disorder
and schizophrenia. Acta Psychiatr Scand 2009;119:192–
198.
77
Psychological Medicine, Page 1 of 4.
doi:10.1017/S0033291710000358
f Cambridge University Press 2010
COMMENTARY
Extended psychosis phenotype – yes:
single continuum – unlikely
A commentary on ‘ Why we need more debate on whether psychotic symptoms lie on a continuum with
normality ’ by David (2010)
N. Kaymaz1,2 and J. van Os1,3*
1
Maastricht University Medical Centre, South Limburg Mental Health Research and Teaching Network, EURON, Maastricht, The Netherlands
Mediant GGZ/Mental Health Care, Postbus 775, 7500 AT Enschede, The Netherlands
3
Division of Psychological Medicine, Institute of Psychiatry, King’s College London, De Crespigny Park, Denmark Hill, London, UK
2
Received 7 February 2010 ; Accepted 8 February 2010
Key words : Diagnosis, psychosis, risk factors.
The accepted dogma is that psychotic disorders come
as diagnosable ‘ things ’, categories, the carriers of
which form a diagnostic boundary, below which reside the healthy non-carriers who do not display the
mental phenomena observed in patients. This model
has come under some pressure, as population research
has shown high rates of psychotic experiences in
people who are not readily diagnosable according to
International Classification of Diseases (ICD) criteria,
Diagnostic and Statistical Manual of Mental Disorders
(DSM) criteria and Research Diagnostic Criteria
(RDC) – suggesting an extended psychosis phenotype.
Research on psychotic experiences in the general
population is in its early stages. A recent meta-analysis
showed that half of the heterogeneity in rates of subclinical psychotic experiences across studies is due to
study cohort and design factors (Linscott & Van Os,
2010). Furthermore, systematic review of the literature
shows that there is evidence not only for a psychometric ‘ continuum ’ (in the sense of an extended
psychosis phenotype blending gradually into clinical
syndromes) (Van Os et al. 2009), but also for an
underlying latent categorical structure of the population (in the sense that regardless of the presence of a
psychometric continuum, the population may still be
composed of two different types of people) (Linscott &
Van Os, 2010).
Therefore, more, and more detailed and finegrained work is necessary. Nevertheless, research on
psychotic experiences suggesting an extended psychosis phenotype already is producing certain ‘ side
* Address for correspondence : J. van Os, Ph.D., Department of
Psychiatry and Psychology, Maastricht University Medical Centre,
PO Box 616 (location DOT12), 6200 MD Maastricht, The Netherlands.
(Email : [email protected])
78
effects ’, essentially to do with perceived blurring
or undermining of diagnostic boundaries, expressed
from a statistical (type I and type II continuum),
philosophical (continuity in risk or essence) or moral
(madness is rare or common) perspective (David,
2010). However, the notion that high rates of psychotic
experiences undermine diagnostic categories only
has relevance if (ICD, DSM, RDC) diagnoses like
‘ schizophrenia ’ represent valid nosological entities,
i.e. ‘ exist ’. As the evidence for the hypothesis of valid
nosological entities is weak, it may be necessary to
formulate an alternative, but scientifically productive,
approach. In this context, it may be helpful to not use
the word ‘ continuum ’, as it carries the implication of a
specific theory, but instead refer to the more agnostic
psychosis ‘ extended phenotype ’ (coined by Howes &
Kapur, 2009).
Extended psychosis phenotype and need for care
Psychotic disorders, spread over a myriad of diagnostic categories in the different classification systems,
in reality represent highly variable syndromal clusters
of continuous psychotic and affective symptom dimensions, in combination with variable degrees of
motivational and neurocognitive impairments (van Os
& Kapur, 2009). Until recently, research on syndromal
clustering of these psychosis dimensions was carried
out almost exclusively in the population of people attending mental health services. The implicit assumption was that symptoms observed in patients with
psychotic disorder naturally did not exist outside the
mental health service. However, now that general
population research is showing that similar clusters
of psychotic, affective, motivational and neurocognitive phenomena are also expressed as extended
N. Kaymaz and J. van Os
Table 1. Factors influencing transition to clinical disorder in representative general population studies of extended psychosis phenotype
1
2
3
4
5
6
7
Factors
References
Number of psychotic experiences
‘ Definite ’ versus ‘ likely ’ psychotic experience
Degree of persistence of psychotic experience over time
Presence of affective dysregulation
Pre-morbid social dysfunction
Presence of negative symptoms
Functional versus ‘ symptomatic ’ coping
Poulton et al. (2000) ; Hanssen et al. (2005) ; Welham et al. (2008)
Poulton et al. (2000)
Dominguez et al. (2009)
van Rossum et al. (2009)
Werbeloff et al. (2009)
Dominguez et al. (in press)
Bak et al. (2003)
Need for care
(b)
Need for care
(a)
Number of psychotic experiences
Number of psychotic experiences
Fig. 1. Simulation of psychosis continuum model (a) versus latent categorical structure model of psychosis (b). On the left side (a)
is depicted the apparent psychosis continuum, greater load of subclinical psychotic experiences predicting greater probability of
clinical outcome. On the right side (b) is depicted a latent categorical structure representing two different groups : one with
psychotic experiences in the context of cognitive and motivational impairments, associated with high probability of need for
care (!) and one group with psychotic experiences of a different origin, associated with low likelihood of need for care ( ).
phenotypes in the general population (Stip &
Letourneau, 2009), a conceptual switch is required : it
may not be productive to consider populations inside
and outside mental health services qualitatively different at the level of symptoms per se, but at the level
of whether or not need for care has developed. One
person with hallucinatory experiences may experience
distress to such a degree that he will visit his GP ;
another may not. The first person may become a
patient with a diagnosis of schizophrenia ; the second
may never receive a diagnosis.
Need for care and the psychopathological,
neurodevelopmental and psychological context
of psychotic experiences
To date, very little research has been carried out
in trying to determine which persons, given a certain
level of expression of the extended psychosis
phenotype, will develop need for care and helpseeking behaviour. Several longitudinal studies in
representative non-help-seeking general population
samples, following individuals with subclinical psychotic experiences over time to determine the likelihood of transition to a clinical outcome, have
produced a number of variables predicting transition
(Table 1). These suggest that not only the ‘ load ’ of
psychotic experiences is important, and what type of
coping the person develops, but also the degree of
persistence over time. Another important observation
is that admixture with affective dysregulation, negative symptoms and pre-morbid social dysfunction also
increases the likelihood of transition to a clinical psychotic outcome. In other words, not just the presence
of psychotic experiences per se, but the psychopathological, developmental and psychological context determines the outcome.
A model for future research
It is generally accepted that the risk for psychotic disorder treated in mental health services is much more
79
Extended psychosis phenotype
prevalent than the rate of psychotic disorder itself ;
more people have risk than there are people with the
disease. Research has demonstrated that (genetic and
non-genetic) risk for psychotic disorder is not behaviourally silent : it is expressed in the area of cognition,
subtle psychotic experiences and sensitivity to stress
(reviewed by van Os & Kapur, 2009). Therefore, the
fact that psychotic experiences are encountered in
non-diagnosed individuals in the general population
is not surprising : psychotic experiences probably represent the behavioural expression of distributed risk
for psychotic disorder. There is also evidence, however, that the causes of positive psychotic symptoms
may only partly overlap with the causes of developmental and cognitive impairment seen in psychotic
disorder (Murray et al. 2004 ; Kaymaz & van Os,
2009b). Therefore, the hypothesis of a single psychosis
continuum lacks explanatory power, and also is not in
agreement with the observation that motivational and
developmental impairments, possibly of genetic origin
(Kaymaz & van Os, 2009a ; Picchioni et al. 2009),
moderate the probability of clinical outcome in those
with subclinical psychotic experiences. Instead, we
suggest that the model depicted in Fig. 1 (simulated
data) may apply. On the one hand, there is an observed apparent continuum (Fig. 1 a), indicating that
the greater the load of subclinical psychotic experiences, the greater the probability of clinical outcome.
This concept is heuristically useful, as it allows for
examination, for example, of how coping and the degree of distress associated with psychotic experiences
may affect development of need for care. On the other
hand, the apparent continuum model may be ‘ confounded ’, at least to a degree, as underneath the continuum, a latent categorical structure may exist,
representing two different groups : one with psychotic
experiences in the context of cognitive and motivational impairments, associated with high probability
of need for care (Fig. 1 b ; black dots) and one group
with psychotic experiences of a different origin, associated with lower likelihood of need for care (Fig. 1 b ;
grey dots). Future research will examine to what degree the continuum model, the latent categorical
structure model, or both (a distinct possibility), apply.
Acknowledgements
N.K. was supported by The Netherlands Organization
for Scientific Research (NWO), project no. 017.002.048.
Declaration of Interest
None.
80
References
Bak M, Myin-Germeys I, Hanssen M, Bijl R, Vollebergh W,
Delespaul P, van Os J (2003). When does experience
of psychosis result in need for care ? A prospective
general population study. Schizophrenia Bulletin 29,
349–358.
David AS (2010). Why we need more debate on whether
psychotic symptoms lie on a continuum with normality.
Psychological Medicine. Published online : 11 February 2010.
doi : 10.1017/S0033291710000188.
Dominguez M, Can Saka M, Lieb R, Wittchen H-U,
Van Os J (in press). Onset of clinical psychosis : do
negative/disorganized symptoms increase risk for positive
symptoms and for impairment ? A 10-year study. American
Journal of Psychiatry.
Dominguez MD, Wichers M, Lieb R, Wittchen HU, van Os J
(2009). Evidence that onset of clinical psychosis is an
outcome of progressively more persistent subclinical
psychotic experiences : an 8-year cohort study.
Schizophrenia Bulletin. Published online : 21 May 2009.
doi : 10.1093/schbul/sbp022.
Hanssen M, Bak M, Bijl R, Vollebergh W, van Os J (2005).
The incidence and outcome of subclinical psychotic
experiences in the general population. British Journal of
Clinical Psychology 44, 181–191.
Howes OD, Kapur S (2009). The dopamine hypothesis of
schizophrenia : version III – the final common pathway.
Schizophrenia Bulletin 35, 549–562.
Kaymaz N, van Os J (2009a). Heritability of structural brain
traits : an endophenotype approach to deconstruct
schizophrenia. International Review of Neurobiology 89,
85–130.
Kaymaz N, van Os J (2009b). Murray et al. (2004) revisited :
is bipolar disorder identical to schizophrenia without
developmental impairment ? Acta Psychiatrica Scandinavica
120, 249–252.
Linscott RJ, Van Os J (2010). Systematic reviews of
categorical versus continuum models in psychosis :
evidence for discontinuous subpopulations underlying a
psychometric continuum. Implications for DSM-V,
DSM-VI, and DSM-VII. Annual Review of Clinical Psychology
6. Published online : 4 January 2010. doi : 10.1146/
annurev.clinpsy.032408.153506.
Murray RM, Sham P, Van Os J, Zanelli J, Cannon M,
McDonald C (2004). A developmental model for
similarities and dissimilarities between schizophrenia and
bipolar disorder. Schizophrenia Research 71, 405–416.
Picchioni MM, Walshe M, Toulopoulou T, McDonald C,
Taylor M, Waters-Metenier S, Bramon E, Regojo A,
Murray RM, Rijsdijk F (2009). Genetic modelling of
childhood social development and personality in twins
and siblings with schizophrenia. Psychological Medicine.
Published online : 29 October 2009. doi : 10.1017/
S0033291709991425.
Poulton R, Caspi A, Moffitt TE, Cannon M, Murray R,
Harrington H (2000). Children’s self-reported psychotic
symptoms and adult schizophreniform disorder : a 15-year
longitudinal study. Archives of General Psychiatry 57,
1053–1058.
N. Kaymaz and J. van Os
Stip E, Letourneau G (2009). Psychotic symptoms as a
continuum between normality and pathology. Canadian
Journal of Psychiatry Revue Canadienne de Psychiatrie 54,
140–151.
Van Os J, Kapur S (2009). Schizophrenia. Lancet 374, 635–645.
Van Os J, Linscott RJ, Myin-Germeys I, Delespaul P,
Krabbendam L (2009). A systematic review and
meta-analysis of the psychosis continuum : evidence for a
psychosis proneness–persistence–impairment model of
psychotic disorder. Psychological Medicine 39, 179–195.
van Rossum I, Dominguez M, Lieb R, Wittchen H-U,
van Os J (2009). Affective dysregulation and reality
distortion : a 10-year prospective study of their association
and clinical relevance. Schizophrenia Bulletin. Published
online : 30 September 2009. doi : 10.1093/schbul/sbp101.
Welham J, Scott J, Williams G, Najman J, Bor W,
O’Callaghan M, McGrath J (2008). Emotional and
behavioural antecedents of young adults who screen
positive for non-affective psychosis : a 21-year birth cohort
study. Psychological Medicine 39, 625–634.
Werbeloff N, Drukker M, Dohrenwend BP, Levav I,
Yoffe R, Van Os J, Davidson M, Weiser M (2009).
Self-reported psychotic symptoms in the community are
associated with increased risk of later hospitalization for
non-affective psychotic disorders (conference abstract).
Schizophrenia Bulletin 35, 74.
81
CHAPTER 3
Affective studies of psychosis
83
Soc Psychiatry Psychiatr Epidemiol (2006) 41:679–685
DOI 10.1007/s00127-006-0086-7
ORIGINAL PAPER
Nil Kaymaz Æ Lydia Krabbendam Æ Ron de Graaf Æ Willem Nolen Æ Margreet ten Have Æ Jim van Os
Evidence that the urban environment specifically impacts on the
psychotic but not the affective dimension of bipolar disorder
Accepted: 10 May 2006 / Published online: 3 July 2006
j Abstract Objectives High rates of psychotic disorders and psychotic symptoms have been found in
urban environments but reports for bipolar affective
illness have been inconsistent, possibly due to failure
to stratify for comorbid psychotic symptoms. It was
hypothesised, therefore, that any effect of urbanicity
on the bipolar phenotype would be moderated by
comorbid psychotic symptoms. Methods In a random, representative population cohort of 7049 adults
with no history of non-affective psychotic disorder,
the cumulative incidence of bipolar and psychotic
symptoms and syndromes, assessed with the CIDI,
was examined over five levels of population density of
place of residence. Similarly, the degree of comorbidity between broadly and narrowly defined bipolar
phenotypes on the one hand, and the dichotomous
presence of broadly (17.2%) and narrowly defined
(3.8%) psychotic symptoms on the other, was examined as a function of population density of place of
residence. Results The rate of bipolar disorder, however defined, was progressively higher in more urbanised areas. However, in models of bipolar
phenotypes, a strong interaction between comorbid
psychosis and level of urbanicity was apparent, indicating that the greater the degree of psychotic comorbidity, the greater the effect size of the urban
environmental factor. For bipolar disorder without
psychosis, no effect of urbanicity was apparent.
Conclusions The results suggest differential environmental causal effects on affective and cognitive
dimensions of bipolar psychopathology that are nevertheless strongly comorbid within the same categorically defined disorder, possibly due to the effect of
shared genetic risk factors.
N. Kaymaz, MD Æ L. Krabbendam, MA, PhD
J. van Os, MD, PhD (&)
Dept. of Psychiatry and Neuropsychology
South Limburg Mental Health Research and Teaching Network
EURON
Maastricht University
P.O. Box 616 (DRT 10)
6200 MD Maastricht, The Netherlands
Tel.: +31-43/3875-443
Fax: +31-43/3875-444
E-Mail: [email protected]
Introduction
R. de Graaf, MA, PhD Æ M. ten Have, MA, PhD
The Netherlands Institute of Mental Health and Addiction
Trimbos Institute
Utrecht, The Netherlands
W. Nolen, MD, PhD
Dept. Psychiatry
University Hospital Groningen
Groningen, The Netherlands
84
There is remarkably consistent evidence that the rate
of psychotic disorder and subclinical psychotic
experiences is higher in urban environments, which
is thought to reflect the impact of an as yet
unidentified environmental causal risk factor [1].
Similar risks, albeit of lower effect size, have been
reported for affective disorder in some studies [2, 3],
but not in others [4]. One likely reason for the
inconsistent results is failure to stratify for comorbid
psychotic symptoms. It is attractive to speculate that
the urban environment will affect the rate of affective disorder in those who also have a propensity to
develop psychosis, but not in those without this
propensity. The importance of such stratification has
been shown in family and high risk studies, where
inconsistent results on the familial association between bipolar disorder and non-affective psychotic
disorder could be retraced in part by failure to
SPPE 86
J. van Os, MD, PhD
Division of Psychological Medicine
Institute of Psychiatry
London, UK
j Key words bipolar disorder – urbanisation –
psychosis
stratify for the presence of psychotic symptoms in
bipolar probands [5–7]. In the current study, we
examined main effects of urbanicity on the rate of
the bipolar phenotype narrowly and broadly defined
[8, 9], and hypothesised that any effect of urbanicity
would be modified synergistically by comorbid psychotic symptoms.
Patients and methods
The Netherlands Mental Health Survey and Incidence Study
(NEMESIS), is a prospective study with three measurement points
over a period of 3 years [10–12]. All procedures were approved by
the local Ethical Committee, and were in accordance with the
Helsinki Declaration of 1975. The current report is based on the
baseline data. A multistage, stratified, random sampling procedure
was used to first select 90 municipalities, then a sample of private
households, and finally a Dutch-speaking individual aged 18–
64 years within each household. Selected households were sent an
introductory letter by the Minister of Health, inviting them to
participate. A total of 7076 individuals provided informed consent
and were interviewed at baseline, representing a response rate of
69.7%. The sample was found to be representative of the Dutch
population in terms of gender, marital status and level of urbanisation [11], with the exception of a slight underrepresentation of
individuals in the age group 18–24 years.
j Instruments
Subjects were interviewed at home. The Composite International
Diagnostic Interview (CIDI) version 1.1 [13–15] was used, yielding
DSM-III-R diagnoses. The CIDI was designed for trained interviewers who are not clinicians and has been found to have high
inter-rater reliability [16, 17], and high test–retest reliability [18–
20]. A total of 90 interviewers, experienced in systematic data
collection, collected the data, having received a 3-day training
course in recruiting and interviewing, followed by a 4-day course at
the WHO-CIDI training centre in Amsterdam. Extensive monitoring and quality checks took place throughout the entire data collection period [11].
j Psychosis ratings
Baseline CIDI lifetime ratings, yielding cumulative incidence figures, from the 17 CIDI core psychosis items on delusions (13 items)
and hallucinations (4 items) were used (items G1-G13, G15, G16,
G20, G21). These concern classic psychotic symptoms involving,
for example, persecution, thought interference, auditory hallucinations and passivity phenomena. All these items can be rated in
six ways: ‘‘1’’—No symptom, ‘‘2’’—Symptom present but not
clinically relevant (not bothered by it and not seeking help for it),
‘‘3’’—Symptom result of ingestion of drugs, ‘‘4’’—Symptom result
of somatic disease, ‘‘5’’—True psychiatric symptom, ‘‘6’’—Symptom may not really be a symptom because there appears to be some
plausible explanation for it. Since psychotic symptoms are difficult
to diagnose in a structured interview [21–23], clinical re-interviews
were conducted over the telephone by an experienced trainee
psychiatrist for all individuals who had at least one rating of ‘‘5’’ or
‘‘6’’, using questions from the Structured Clinical Interview for
DSM-III-R (SCID), an instrument with proven reliability and
validity in diagnosing schizophrenia [24]. The proportion of eligible individuals that was successfully re-interviewed by the clinician was 47.2%. Whether or not eligible individuals were reinterviewed was not associated with level of urbanicity (OR = 0.9,
95% CI: 0.8, 1.02, P = 0.10; proportions in lowest and highest levels
of urbanicity, respectively, re-interviewed: 13.9% and 15.4%; not re-
interviewed: 15.5% and 19.8%). CIDI ratings were corrected on the
basis of these clinical interviews, and the NEMESIS DSM-III-R
diagnoses of psychotic disorder are based on the data from these
clinical re-interviews.
Non-affective psychotic disorder was defined as any DSM-III-R
non-affective psychotic diagnosis (n = 26, 0.4%). A broad measure
of psychosis was defined as any CIDI rating of 2,3,4,5 or 6 on any of
the 17 CIDI core psychosis items. In a previous study, it was shown
that all these 6 different ratings on the CIDI psychosis items were
strongly associated with each other, including the clinical reinterview ratings of psychotic symptom (i.e. a rating of ‘‘5’’ on any
of the CIDI psychosis items). In addition, the 6 different ratings
independently showed a similar pattern of associations with known
risk factors for psychosis [25, 26]. As they, therefore, appear to
reflect the same underlying latent dimension of ‘‘psychosis’’ they
were joined together into a single broad rating of psychotic experience for the purpose of the current study (hereafter: ‘‘psychosis
broadly defined’’). In order to check on the validity of this procedure, associations were also examined with clinical psychotic
symptoms, more narrowly defined as a clinical re-interview rating
of psychosis (i.e. a rating of ‘‘5’’ on any of the CIDI psychosis items,
hereafter: ‘‘psychosis narrowly defined’’).
j Bipolar phenotypes
At baseline, a CIDI lifetime diagnosis of bipolar disorder,
reflecting cumulative incidence, was made in 132 individuals
(1.9%). A mania symptom score was constructed using ratings
from the 11 items of the CIDI core mania section (section F; see
[27]). The mania symptom score consisted of the sum score of
these 11 items. The range of this score was 0–10 symptoms, and
the minimum positive score was 3 symptoms (absence of score of
1 or 2 due to cut-off stem questions in the CIDI). Apart from the
CIDI bipolar diagnosis, two broader bipolar outcomes, in line
with current conceptualisation of the bipolar phenotype [8, 9]
were used, derived from this score: a score of at least 3 symptoms
(hereafter: mania 3-symptoms) and a score of at least 5 symptoms
(hereafter: mania 5-symptoms). For the two broader phenotypes,
manic rather than depressive symptoms were used, given the fact
that mania is the distinguishing feature between unipolar and
bipolar affective phenotypes and no accepted method exists for
defining subclinical bipolar disorder using both depressive and
manic symptoms.
Level of urbanicity
Five levels of urbanisation were defined, following the standard
classification of urbanisation of place of residence according to the
Dutch Central Bureau of Statistics. These are based on the density
of addresses per km2 in an area and are classified as <500, 500–999,
1000–1499, 1500–2499 and ‡2500. This density is calculated by
assessing, for each address, the density of addresses in a circle of
1 km around that address. The density of addresses in an area is
then calculated as the mean address density of all the addresses in
that area [28].
j Data analyses
Main effect of urbanicity
The cumulative incidences of bipolar disorder, 3-symptom mania
and 5-symptom mania were examined in relation to level of
urbanicity of place of residence, adjusted for the a priori selected
possible confounding effects of age in years, sex, level of education
(4-levels) and cannabis use [26, 29, 30].
Logistic regression yielding odds ratios and 95% confidence
intervals [31] was used to examine associations between bipolar
outcomes and level of urbanicity.
85
Interactions
The classic problem with regard to possible co-participation
between causes in nature (biological synergism) is how such
synergism can be inferred from statistical manipulations with
research data (statistical interaction), in particular with regard to
the choice of additive (change in risk occurs by adding a
quantity) or multiplicative (change in risk occurs by multiplying
with a quantity) models. Recent progress in the study of interactions indicates that the most frequently used statistical models
of interaction are not suitable to identify biological synergism.
For example, the commonly used statistical models in which two
causes multiply each other’s effects (multiplicative models) assume that individuals who are exposed to both risk factor A and
risk factor B could not have become ill because the effect of A
alone or B alone [32]. It has recently been shown that the true
degree of biological synergism can be better estimated from (but
is not the same as) the additive statistical interaction (see [32]).
This method was recently applied to schizophrenia showing
synergy between traumatic head injury and familial liability [33],
between cannabis and psychosis liability [34] and between
urbanicity and familial morbid risk for psychosis [35].
In order to calculate the statistical interaction under an
additive model, the BINREG procedure in the STATA statistical
programme [36], which fits generalised linear models for the
binomial family estimating risk differences, was used to model
interactions.
Table 1 Cumulative incidences of mania 3-symptom, mania 5-symptom and
bipolar disorder by 5 levels of population density
Population density
M3a (%)
M5a (%)
bipa (%)
1
2
3
4
5
Total
OR linear trendb
(95% CI)
P
Adjustedc OR
linear trend
(95% CI)
P
2.8
2.7
3.8
5.0
6.1
4.0
1.27
(1.16, 1.39)
<0.001
1.22
1.4
0.9
2.1
2.9
4.0
2.2
1.43
(1.26, 1.62)
<0.001
1.34
1.3
1.2
2.2
2.1
2.8
1.9
1.24
(1.09, 1.42)
0.001
1.18
(1.11, 1.34)
<0.001
(1.18, 1.53)
<0.001
(1.03, 1.35)
0.016
a
‘‘M3’’ is broad mania phenotype of at least 3 manic symptoms, ‘‘M5’’ is broad
mania phenotype of at least 5 manic symptoms, ‘‘bip’’ is DSM-III-R bipolar
disorder
b
The summary increase in risk, expressed as odds ratio, associated with one
unit change in the five-level exposure
c
Adjusted for age in years, sex, level of education (4-levels) and cannabis use
Table 2 Cumulative incidences of mania 3-symptom, mania 5-symptom and
bipolar disorder by 5 levels of population density, stratified by presence of
psychosis broadly defined
Risk set
Population density
The risk set consisted of all individuals who had participated in
the baseline CIDI interview, with the exception of the 26 individuals with a lifetime diagnosis of non-affective psychotic disorder (n = 26), resulting in a sample for analysis of 7049
individuals.
M3a (%)
No psychotic symptom
1
1.7
2
1.4
3
2.1
4
2.0
5
2.5
Total
1.9
b
0.22
RD linear trend (%)
(95% CI)
(0.00, 0.48)
P
0.095
At least one psychotic symptom
1
9.9
2
10.6
3
12.2
4
16.7
5
18.3
Total
14.1
2.3
RD linear trendb (%)
(95% CI)
(0.92, 3.7)
P
0.001
Interactionc
2
(v , P, df = 1)
8.4, 0.004
Results
j Sample description
The sample included 3287 men (47%) and 3762
women (53%) and the mean age was 41.2 years
(SD = 12.2). The cumulative incidence of psychosis
broadly defined was 17.2% (n = 1211) and the
cumulative incidence of psychosis narrowly defined
was 3.8% (n = 267). The cumulative incidence of
mania 3-symptoms was 4.0% (n = 283), of mania 5symptoms was 2.3% (n = 161) and of DSM-III-R
bipolar disorder 1.9% (n = 132).
M5a (%)
0.8
0.4
0.9
0.8
1.6
0.9
0.16
(0.00, 0.33)
0.076
bipa (%)
0.8
0.7
1.2
0.9
1.5
1.0
0.16
(0.00, 0.34)
0.10
4.9
4.2
7.9
11.3
12.2
8.7
2.1
(1.1, 3.2)
<0.001
4.3
4.6
7.1
6.7
7.2
6.2
0.78
(0.00, 1.8)
0.12
13.0, 0.0003
1.5, 0.21
a
j Main effects urbanicity and psychosis
comorbidity
The five level urbanicity rating was strongly associated with the rate of mania, however defined. Both the
CIDI bipolar diagnosis, as the two broader bipolar
outcomes were all progressively more common in
more urbanised areas (Table 1). Adjustment for the a
priori selected possible confounding effects of age,
sex, level of education (4-level) only changed the
parameters by a small amount (Table 1).
86
‘‘M3’’ is broad mania phenotype of at least 3 manic symptoms, ‘‘M5’’ is broad
mania phenotype of at least 5 manic symptoms, ‘‘bip’’ is DSM-III-R bipolar
disorder
b
The summary increase in risk, expressed as risk difference, associated with
one unit change in the five-level exposure
c
Tests whether RD linear trend for mania with psychosis is significantly different from RD linear trend for mania without psychosis
j Urbanicity and psychosis interaction effects
There was a directionally positive interaction on the
additive scale between urbanicity and broadly defined
psychosis in the models of mania, reaching statistical
significance for 2 of the 3 mania outcomes (mania 3-
Table 3 Cumulative incidences of mania 3-symptom, mania 5-symptom and
bipolar disorder by 5 levels of population density, stratified by presence of
psychosis narrowly defined
Population density
No psychotic symptom
1
2
3
4
5
Total
RD linear trendb (%)
(95% CI)
P
At least one
psychotic symptom
1
2
3
4
5
Total
RD linear trendb (%)
(95% CI)
P
Interactionc
(v2, P, df = 1)
M3a (%)
M5a (%)
2.6
2.0
3.3
3.5
4.4
3.1
0.50
(0.19, 0.81)
0.001
1.1
0.6
1.7
1.8
2.7
1.5
0.39
(0.18, 0.60)
<0.001
11.1
28.2
14.8
30.4
36.5
26.0
5.4
(0.69, 9.6)
0.024
6.3, 0.01
11.1
15.4
11.5
22.8
27.0
19.0
3.9
(0.63, 7.4)
0.020
4.4, 0.04
bipa (%)
1.2
0.8
1.7
1.6
2.2
1.5
0.27
(0.00, 0.48)
0.012
3.7
15.4
13.1
10.1
12.7
11.5
0.82
()0.02, 3.8)
0.59
0.13, 0.72
a
‘‘M3’’ is broad mania phenotype of at least 3 manic symptoms, ‘‘M5’’ is broad
mania phenotype of at least 5 manic symptoms, ‘‘bip’’ is DSM-III-R bipolar
disorder
b
The summary increase in risk, expressed as risk difference, associated with
one unit change in the five-level exposure
c
Tests whether RD linear trend for mania with psychosis is significantly different from RD linear trend for mania without psychosis
symptoms: v2 = 8.4, P = 0.004, mania 5-symptoms:
v2 = 13.0, P = 0.0003; DSM-III-R bipolar disorder:
v2 = 1.5, P = 0.21) (Table 2). Similar results were
apparent for interactions between urbanicity and
psychosis narrowly defined (mania 3-symptoms:
v2 = 6.3, P = 0.01, mania 5-symptoms: v2 = 4.4,
P = 0.04; DSM-III-R bipolar disorder: v2 = 0.13,
P = 0.72) (Table 3). Stratified analyses revealed that
in mania outcomes without comorbid psychosis narrowly defined, only a weak effect of urbanicity remained (Table 3), whereas in mania outcomes
without comorbid psychosis broadly defined the effect of urbanicity disappeared altogether (Table 2).
For bipolar disorder without psychosis, no strong
or clear effect of urbanicity was apparent. Thus, in the
analyses of bipolar disorder where only narrowly
defined comorbid psychosis had been excluded, a
weak effect of urbanicity was still apparent. However,
for bipolar disorder outcomes where all psychosis had
been excluded, an (inconclusive) effect of urbanicity
was only possibly apparent for the highest category of
urbanicity and the broadest definition of mania. Thus,
the conservative conclusion is that although a weak
association between mania without psychosis and
urbanicity cannot be excluded, by far the greatest part
of the apparent association between bipolar phenotypes and urbanicity is due to comorbid psychosis.
Indeed, although the incidence of other psychiatric
disorders may also be higher in urban areas [3, 4], the
effect size for psychotic disorders appears to be much
larger, possibly explaining around 30% of all schizophrenia incidence [1, 4, 44]. Given the fact that
associations with urbanicity are thought to reflect the
impact of an environmental exposure that interacts
with genetic liability to produce illness [1, 35, 37],
these results should be interpreted in the light of
previous work in the field of both molecular genetics
and genetic epidemiology that has shown that there is
substantial sharing of genetic risk between bipolar
disorder and non-affective psychosis [38–40]. Thus, if
two individuals have a similar amount of shared genetic liability for both bipolar disorder and schizophrenia, the one that becomes subsequently exposed
to an urban environment may develop a more ‘‘psychotic’’ illness, whereas the one not exposed to an
urban environment may develop a more ‘‘pure’’
manic illness, suggesting that within the bipolar
spectrum, the impact of the urban environment on
the occurrence of more ‘‘psychotic’’ illness may be
mediated by a different pathway than the rate of more
‘‘pure’’ manic illness. Research suggests that this type
of gene–environment interactions may not be
uncommon. For example, a similar scenario has been
described for anxiety and depression, two conditions
that have been shown to share genes that, however,
produce differential outcomes depending on subsequent exposure to divergent environmental risk
factors, some resulting in anxiety outcomes, and
others in depression [41–43].
Discussion
The main finding of this study was that in models of
bipolar phenotypes, a strong interaction between
psychosis and level of urbanicity was apparent, indicating that the greater the degree of psychotic comorbidity, the greater the effect size of the urban
environmental factor. Although the interaction term
was not significant for DSM-III-R bipolar disorder,
the difference in urbanicity effect size in the groups
with and without comorbid psychosis was equally
large so that lack of power is the likely explanation for
the inconclusive interaction.
j Possible mechanisms
Recent studies have shown that the impact of the
environmental factors that are thought to mediate the
effect of urbanicity is conditional on pre-existing
indicators of genetic risk for psychosis [35, 37, 45].
However, a plausible mechanism explaining the
pathway from impact of the exposure to onset of
psychotic symptoms is lacking, as it is unclear which
underlying factors in urban areas are responsible for
the causation of psychotic symptoms in genetically
87
pre-disposed individuals [46]. Recent studies, focussing on within-city neighbourhood contrasts, have
demonstrated that the impact of environmental factors in the urban area varies across neighbourhoods
in such a way that the increase in the incidence of
psychosis or psychosis-like phenomena is likely not
associated with indicators of material deprivation, but
with indicators of social isolation and social cohesion
[47]. These studies have focused on the developmental effects of variables referred to as ‘‘social capital’’ and suggest that in particular cognitive social
capital, or the degree of perceived mutual trust,
bonding, safety and social control in neighbourhoods,
exerts a developmental impact on the mental health of
the children growing up in these environments [48–
51]. The hypothesised effects of cognitive social capital can be readily integrated in current hypotheses of
cognitive neuropsychiatric models of psychotic
symptoms [52, 53].
HOTDECK procedure. This revealed that the interaction term effect size decreased only slightly for the
mania 5-symptom model (12% reduction effect size)
whereas it increased substantially for the mania 3symptom model (47% increase effect size). These
results therefore, suggest that bias due to differential
re-interview rate did not occur.
Another bias could be the examination of lifetime
rates of psychotic disorder in relation to current urban residence. For example, symptomatic subjects
could have ‘drifted’ to urban areas. However, in a
previous study in the Netherlands we found that there
was a high degree of lifetime stability of urban
exposure status, indicating that current exposure is
likely to reflect stable lifetime exposure in most cases.
Furthermore, work from Denmark has suggested drift
is an unlikely explanation for the urbanicity effect,
since changing exposure status in childhood changes
adult risk for schizophrenia, indicating causality rather than drift [58].
j Methodological issues
These results should be considered in the light of
several limitations: Psychotic symptom ratings were
assessed by lay interviewers (CIDI ratings of 1–4),
whereas the other CIDI ratings [5, 6] were assessed by
clinicians through telephone interviews in approximately 50% of eligible cases. It is likely that with lay
interviewer ratings a degree of misclassification did
occur. However, the findings with both broad (only
lay interviewer based) and narrow (also clinical reinterview based) psychosis concurred, suggesting that
lay interviews were reasonably accurate, something
which was previously reported [54].
Bias could also have been introduced by differential rating of psychosis due to incomplete clinical
re-interview rates at baseline. This is unlikely, however, as the probability of re-interview was not
associated with the exposure, urbanicity, which
would have been required for bias to occur [55]. In
order to examine the possibility of bias further,
sensitivity analyses were conducted to examine
whether differential re-interview rates could have
biased the findings. This was done by setting values
of psychosis narrowly defined to missing if individuals had been eligible for re-interview by clinicians
but had not been re-interviewed (n = 253). Subsequently, multiple imputation of missing values [56,
57] of psychosis narrowly defined was applied using
the HOTDECK command in STATA. The HOTDECK
procedure is used several times within a multiple
imputation sequence since missing data are imputed
stochastically
rather
than
deterministically.
Five imputation sequences were run, yielding 5 data
sets in which the average coefficient of the urbanicity · psychosis narrowly defined interaction term
in the model of the two mania outcomes, which had
shown significant interactions (mania 3-symptom
and mania 5-symptom) was estimated within the
88
j Acknowledgement
Funded by the Dutch Department of Health.
References
1. Van Os J (2004) Does the urban environment cause psychosis?
Br J Psychiat 184(4):287–288
2. Kessler RC, Rubinow DR, Holmes C, Abelson JM, Zhao S (1997)
The epidemiology of DSM-III-R bipolar I disorder in a general
population survey. Psychol Med 27(5):1079–1089
3. Sundquist K, Frank G, Sundquist J (2004) Urbanisation and
incidence of psychosis and depression: follow-up study of
4.4 million women and men in Sweden. Br J Psychiat
184(4):293–298
4. Mortensen PB, Pedersen CB, Melbye M, Mors O, Ewald H
(2003) Individual and familial risk factors for bipolar affective disorders in Denmark. Arch Gen Psychiat 60(12):1209–
1215
5. Kendler KS, McGuire M, Gruenberg AM, Spellman M, O’Hare
A, Walsh D (1993) The Roscommon Family Study. II. The risk
of nonschizophrenic nonaffective psychoses in relatives. Arch
Gen Psychiat 50(8):645–652
6. Erlenmeyer Kimling L, Adamo UH, Rock D, Roberts SA, Bassett
AS, Squires Wheeler E, et al. (1997) The New York High-Risk
Project. Prevalence and comorbidity of axis I disorders in offspring of schizophrenic parents at 25-year follow-up. Arch Gen
Psychiat 54(12):1096–1102
7. Kendler KS, Gruenberg AM, Tsuang MT (1986) A DSM-III
family study of the nonschizophrenic psychotic disorders. Am J
Psychiat 143(9):1098–1105
8. Judd LL, Akiskal HS (2003) The prevalence and disability of
bipolar spectrum disorders in the US population: re-analysis of
the ECA database taking into account subthreshold cases. J
Affect Disord 73(1–2):123–131
9. Angst J, Gamma A, Benazzi F, Ajdacic V, Eich D, Rossler W
(2003) Toward a re-definition of subthreshold bipolarity:
epidemiology and proposed criteria for bipolar-II, minor
bipolar disorders and hypomania. J Affect Disord 73(1–
2):133–146
10. Bijl RV, Ravelli A, Van Zessen G (1998) Prevalence of psychiatric disorder in the general population: results from the
Netherlands mental health survey and incidence study. Soc
Psychiat Psychiat Epidemiol 33:587–596
11. Bijl RV, Van Zessen G, Ravelli A, De Rijk C, Langendoen Y
(1998) The Netherlands mental health survey and incidence
study (NEMESIS): objectives and design. Soc Psychiat Psychiat
Epidemiol 33:581–587
12. Alegria M, Bijl RV, Lin E, Walters EE, Kessler RC (2000)
Income differences in persons seeking outpatient treatment
for mental disorders: a comparison of the United States with
Ontario and The Netherlands. Arch Gen Psychiat 57(4):383–
391
13. Robins LN, Wing J, Wittchen HU, Helzer JE, Babor TF, Burke J,
et al. (1988) The Composite International Diagnostic Interview.
An epidemiologic Instrument suitable for use in conjunction
with different diagnostic systems and in different cultures. Arch
Gen Psychiat 45(12):1069–1077
14. World Health Organisation (1990) Composite International
Diagnostic Interview (CIDI) Version 1.0. World Health Organisation, Geneva
15. Smeets RMW, Dingemans PMAJ (1993) Composite International Diagnostic Interview (CIDI) Versie 1.1. World Health
Organisation, Amsterdam/Geneva
16. Wittchen HU, Robins LN, Cottler LB, Sartorius N, Burke JD,
Regier D (1991) Cross-cultural feasibility, reliability and sources of variance of the Composite International Diagnostic
Interview (CIDI). The Multicentre WHO/ADAMHA Field Trials
[published erratum appears in Br J Psychiatry 1992 Jan;
160:136]. Br J Psychiat 159:645–653
17. Cottler LB, Robins LN, Grant BF, Blaine J, Towle LH, Wittchen
H-U, et al. (1991) The CIDI-core substance abuse and dependence questions: cross-cultural and nosological issues. Br J
Psychiat 159:653–658
18. Semler G, Von Cranach M, Wittchen H-U (1987) Comparison
between the Composite International Diagnostic Interview and
the present state examination. Report to the WHO/ADAMHA
task force on instrument development. World Health Organisation, Geneva
19. Wacker HR, Battegay R, Mullejans R, Schlosser C (1990) Using
the CIDI in the general population. In: Stefanis CN, Rabavilas
AD, Soldatos CR (eds) Psychiatry: a world perspective. Elsevier
Science Publishers, Amsterdam, pp 138–143
20. Wittchen HU (1994) Reliability and validity studies of the WHO
– Composite International Diagnostic Interview (CIDI): a critical review. J Psychiat Res 28(1):57–84
21. Helzer JE, Robins LN, McEvoy LT, Spitznagel EL, Stoltzman
RK, Farmer A, et al. (1985) A comparison of clinical and
diagnostic interview schedule diagnoses. Physician reexamination of lay-interviewed cases in the general population. Arch
Gen Psychiat 42(7):657–666
22. Anthony JC, Folstein M, Romanoski AJ, Von Korff MR,
Nestadt GR, Chahal R, et al. (1985) Comparison of the lay
diagnostic interview schedule and a standardised psychiatric
diagnosis. Experience in Eastern Baltimore. Arch Gen Psychiat 42:667–675
23. Cooper SA, Collacott RA (1994) Relapse of depression in people
with Down’s syndrome. Br J Dev Disabil 40(78, Pt 1):32–37
24. Spitzer RL, Williams JB, Gibbon M, First MB (1992) The
Structured Clinical Interview for DSM-III-R (SCID). I: history,
rationale, and description. Arch Gen Psychiat 49(8):624–629
25. Van Os J, Hanssen M, Bijl R-V, Ravelli A (2000) Straus (1969)
revisited: a psychosis continuum in the general population?
Schizophr Res 45(1–2):11–20
26. Van Os J, Hanssen M, Bijl RV, Vollebergh W (2001) Prevalence
of psychotic disorder and community level of psychotic
symptoms: an urban–rural comparison. Arch Gen Psychiat
58(7):663–668
27. Krabbendam L, Myin-Germeys I, Graaf Rd, Vollebergh W,
Nolen WA, Os Jv (2004) Dimensions of depression, mania, and
psychosis in the general population. Psychol Med 34(7):1177–
1186
28. Central Bureau of Statistics (1993) Bevolking der Gemeenten
van Nederland. CBS-Publications, The Hague
29. Lewis G, David A, Andreasson S, Allebeck P (1992) Schizophrenia and city life. Lancet 340(8812):137–140
30. Marcelis M, Navarro-Mateu F, Murray R, Selten JP, Van Os J
(1998) Urbanization and psychosis: a study of 1942–1978 birth
cohorts in The Netherlands. Psychol Med 28(4):871–879
31. Platt RW (1997) Logistic regression and odds ratios. Inj Prev
3(4):294
32. Darroch J (1997) Biologic synergism and parallelism [see
comments]. Am J Epidemiol 145(7):661–668
33. Corcoran C, Malaspina D (2001) Traumatic brain injury and
schizophrenia. Int J Mental Health 30:17–33
34. Van Os J, Bak M, Hanssen M, Bijl RV, de Graaf R, Verdoux H
(2002) Cannabis use and psychosis: a longitudinal populationbased study. Am J Epidemiol 156(4):319–327
35. Van Os J, Hanssen M, Bak M, Bijl RV, Vollebergh W (2003) Do
urbanicity and familial liability coparticipate in causing psychosis? Am J Psychiat 160(3):477–482
36. StataCorp (2001) STATA Statisctical Software: Release 7.0.
College Station, Texas
37. Van Os J, Pedersen CB, Mortensen PB (2004) Confirmation of
synergy between urbanicity and familial liability in the causation of psychosis. Am J Psychiat 161(12):2312–2314
38. Cardno AG, Rijsdijk FV, Sham PC, Murray RM, McGuffin P
(2002) A twin study of genetic relationships between psychotic
symptoms. Am J Psychiat 159(4):539–545
39. Berrettini W (2004) Bipolar disorder and schizophrenia: convergent molecular data. Neuromol Med 5(1):109–117
40. Berrettini W (2003) Evidence for shared susceptibility in
bipolar disorder and schizophrenia. Am J Med Genet
123C(1):59–64
41. Kendler KS, Heath AC, Martin NG, Eaves LJ (1987) Symptoms
of anxiety and symptoms of depression. Same genes, different
environments? Arch Gen Psychiat 44(5):451–457
42. Kendler KS, Neale MC, Kessler RC, Heath AC, Eaves LJ (1992)
Major depression and generalized anxiety disorder. Same
genes, (partly) different environments? Arch Gen Psychiat
49(9):716–722
43. Kendler KS (1996) Major depression and generalised anxiety
disorder. Same genes, (partly) different environments – revisited. Br J Psychiat Suppl (30):68–75
44. Mortensen PB, Pedersen CB, Westergaard T, et al. (1999) Effects of family history and place and season of birth on the risk
of schizophrenia. N Engl J Med 340(8):603–608
45. Spauwen J, Krabbendam L, Roselind L, Wittchen HU, van Os J
(2006) Evidence that the outcome of developmental expression
of psychosis is worse for adolescents growing up in an urban
environment. Psychol Med 36(3):407–415
46. Krabbendam L, van Os J (2005) Schizophrenia and urbanicity: a
major environmental influence – conditional on genetic risk.
Schizophr Bull 31(4):795–799
47. Van Os J, Driessen G, Gunther N, Delespaul P (2000) Neighbourhood variation in incidence of schizophrenia: evidence for
person–environment interaction. Brit J Psychiat 176:243–248
48. McKenzie K, Whitley R, Weich S (2002) Social capital and
mental health. Brit J Psychatry 181:280–283
49. De Silva MJ, McKenzie K, Harpham T, Huttly SR (2005) Social
capital and mental illness: a systematic review. J Epidemiol
Commun Health 59(8):619–627
50. Drukker M, Kaplan C, Feron F, van Os J (2003) Children’s
health related quality of life, neighbourhood socio-economic
deprivation and social capital: a contextual analysis. Soc Sci
Med 57(5):825–841
51. Drukker M, Buka SL, Kaplan C, McKenzie K, van Os J (2005)
Social capital and young adolescents, perceived health in different sociocultural settings. Soc Sci Med 61(1):185–198
52. Blackwood NJ, Howard RJ, Bentall RP, Murray RM (2001)
Cognitive neuropsychiatric models of persecutory delusions.
Am J Psychiat 158(4):527–539
53. Garety PA, Kuipers E, Fowler D, Freeman D, Bebbington PE
(2001) A Cognitive model of the positive symptoms of psychosis. Psychol Med 31(2):189–195
54. Bak M, Delespaul P, Hanssen M, de Graaf R, Vollebergh W, van
Os J (2003) How false are ‘‘false’’ positive psychotic symptoms?
Schizophr Res 62(1–2):187–189
89
55. Hennekens CH, Buring JE (1987) Analysis of epidemiologic
studies: evaluating the role of bias. In: Hennekens CH, Buring
JE (eds) Epidemiology in medicine. Little, Brown and Company, Toronto, pp 272–287
56. Hopke PK, Liu C, Rubin DB (2001) Multiple imputation for
multivariate data with missing and below-threshold measurements: time-series concentrations of pollutants in the Arctic.
Biometrics 57(1):22–33
90
57. Rubin DB, Schenker N (1991) Multiple imputation in healthcare databases: an overview and some applications. Stat Med
10(4):585–598
58. Pedersen CB, Mortensen PB (2001) Evidence of a dose–
response relationship between urbanicity during upbringing
and schizophrenia risk. Arch Gen Psychiat 58(11):1039–
1046
Journal of Affective Disorders 98 (2007) 55 – 64
www.elsevier.com/locate/jad
Research report
The impact of subclinical psychosis on the transition from
subclinicial mania to bipolar disorder
Nil Kaymaz a , Jim van Os a,b,⁎, Ron de Graaf c , Margreet ten Have c ,
Willem Nolen d , Lydia Krabbendam a
a
Department of Psychiatry and Neuropsychology, South Limburg Mental Health Research and Teaching Network, EURON,
Maastricht University, PO BOX 616 (DRT 10), 6200 MD Maastricht, The Netherlands
b
Division of Psychological Medicine, Institute of Psychiatry, De Crespigny Park, Denmark Hill, London SE5 8AF, UK
c
The Netherlands Institute of Mental Health and Addiction (Trimbos Institute), Utrecht, The Netherlands
d
Department of Psychiatry, University Medical Center Groningen, PO BOX 30.001, 9700 RB Groningen, The Netherlands
Received 25 April 2006; received in revised form 29 June 2006; accepted 30 June 2006
Available online 28 August 2006
Abstract
Background: In the general population, symptoms of mania and psychosis are more broadly distributed than their associated clinical
syndromes. Little is known, however, about how these subclinical population phenotypes co-vary with and impact on each other.
Method: In a representative population cohort of 7076 adults, prevalence of mania and psychosis symptoms and syndromes were
assessed with the CIDI at baseline, at one (T1) and two years later (T2). The degree of comorbidity between subclinical mania and
subclinical psychosis was examined, as well as the impact of subclinical comorbidity on social impairment and transition from
subclinical mania to onset of bipolar disorder.
Results: The lifetime prevalence of at least one manic and one psychotic symptom was 4.1% and 4.2% respectively. Excluding
individuals with any lifetime DSM-III-R bipolar or psychotic disorder (n = 218), these prevalences were 2.3% (subclinical mania)
and 2.8% (subclinical psychosis). Individuals with subclinical mania had a 17% risk of subclinical psychosis, compared with 2.3%
in those without (P b 0.000). Comorbid subclinical psychosis in individuals with subclinical mania was much more predictive of a
future diagnosis of bipolar disorder (positive predictive values of 3% versus 10% respectively).
Conclusion: Subclinical phenotypes of mania and psychosis are more prevalent than their clinical counterparts and cluster together.
The risk factors for psychosis may facilitate the formation of more “toxic” combinations of subclinical mania and subclinical
psychosis with a higher probability of transition to bipolar disorder. A better understanding of this pathway is crucial for the
development of early interventions.
© 2006 Elsevier B.V. All rights reserved.
Keywords: Mania; Psychosis; Genetics; Environment; Prevention; Comorbidity
1. Introduction
⁎ Corresponding author. Department of Psychiatry and Neuropsychology, Maastricht University, PO BOX 616 (DRT 10), 6200 MD
Maastricht, The Netherlands. Tel.: +31 43 3875443; fax: +31 43
3875444.
E-mail address: [email protected] (J. van Os).
It is increasingly recognised that in the general population, the expression of mania and psychosis is more
broadly distributed than the associated clinical syndromes (Akiskal et al., 2000; Angst, 2002; Johns and van
Os, 2001; Judd et al., 2003; Peters et al., 1999; Stefanis et
0165-0327/$ - see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.jad.2006.06.028
91
N. Kaymaz et al. / Journal of Affective Disorders 98 (2007) 55–64
al., 2002; van Os et al., 2001). Thus, even though the
prevalence of a clinical disorder may be low, the prevalence of isolated symptoms can be much higher. In
addition to phenomenological correspondence, continuity between the clinical and subclinical phenotypes is
suggested by similar associations with demographic factors (Johns and van Os, 2001; van Os et al., 2000a;
Verdoux et al., 1998), transitions over time from the
subclinical to the clinical phenotype (Chapman et al.,
1994; Hanssen et al., 2005; Poulton et al., 2000), familial
co-clustering of the clinical and the subclinical phenotype (Kendler et al., 1993), sharing of cognitive and
motor deficits (Lenzenweger, 1994; Matsui et al., 2004;
Neumann and Walker, 2003; Voglmaier et al., 2000), and
sharing of risk genes (Stefanis et al., 2004a).
There is substantial overlap between dimensions of
mania and psychosis in clinical samples, which may be
due to the sharing of genetic and non-genetic aetiological
influences. Familial co-aggregation suggests that psychosis shares common genetic risk factors with mania
(Bailer et al., 2002; Cardno et al., 2002; Kendler and
Gardner, 1997; Potash et al., 2001; Valles et al., 2000)
and recent studies have indeed implied shared susceptibility genes (Bramon and Sham, 2001; Green et al.,
2005; Maier et al., 2005; Walss-Bass et al., 2005). Although there are quantitative differences in effect size
(Van Os et al., 1998), risk factors such as life events,
ethnic group, prenatal famine and urban birth also tend to
overlap to a degree between psychotic and mood disorders (Bebbington et al., 1993; Brown et al., 2000;
Marcelis et al., 1998; van Os, 1996) as do cognitive
impairments and high levels of neuroticism prior to the
onset of disorder (Goodwin et al., 2003; Jones et al.,
1994; Myin-Germeys et al., 2002; Van Os and Jones,
2001b).
Given the continuity between clinical and subclinical
levels of psychopathology within a particular symptom
domain, it can be surmised that the overlap between
mania and psychosis should also be detectable at lower
levels of the continuum. However, little is known about
how subclinical population phenotypes of mania and
psychosis co-vary with and have impact on each other
and which variables affect observed comorbidity between subclinical phenotypes. It was hypothesised that
the demographic, familial, psychological, and social risk
factors for clinical psychotic disorder (Van Os, 2000;
Van Os et al., 2000b; Van Os et al., 2005) would also
predict the presence of subclinical psychosis in individuals with subclinical mania. It was further hypothesised that since the presence of psychosis in clinical
bipolar disorder predicts poorer outcome (Coryell et al.,
2001), the coexistence of psychotic features with sub-
92
clinical mania would similarly have a negative impact on
social impairment and be predictive of a future diagnosis
of bipolar disorder.
2. Methods
The Netherlands Mental Health Survey and Incidence
Study (NEMESIS), is a prospective study with three
measurement points over a period of 3 years (Alegria et
al., 2000; Bijl et al., 1998a,b). The current report is based
on the baseline data, with the exception of the predictive
analysis of the post-baseline diagnosis of bipolar disorder. A multistage, stratified, random sampling procedure
was used to first select 90 municipalities, then a sample
of private households, and finally a Dutch-speaking
individual aged 18–64 years within each household.
Selected households were sent an introductory letter by
the Minister of Health, inviting them to participate. A
total of 7076 individuals provided informed consent and
was interviewed at baseline, representing a response rate
of 69.7%. Nearly 44% of non-responders agreed to fill in
a postal questionnaire, including a General Health
Questionnaire (Goldberg and Williams, 1988), and
were found to have the same mean GHQ score (responders: 1.19; non-responders: 1.16). At T1, 5618 subjects
participated for the second time; at T2, 4848 subjects
participated. The sample was found to be representative
of the Dutch population in terms of sex, marital status
and level of urbanisation (Bijl et al., 1998b), with the
exception of a slight underrepresentation of individuals
in the age group 18–24 years. As this was a study of
relative rather than absolute risk, no poststratification
weightings were applied to the data.
2.1. Instruments
Subjects were interviewed at home. The Composite
International Diagnostic Interview (CIDI; http://www3.
who.int/cidi/) version 1.1 (Robins et al., 1988; Smeets
and Dingemans, 1993) was used, yielding DSM-III-R
diagnoses. The CIDI was designed for trained interviewers who are not clinicians and has been found to
have high inter-rater reliability (Cottler, 1991; Wittchen
et al., 1991), and high test–retest reliability (Semler et
al., 1987; Wacker et al., 1990; Wittchen, 1994). Ninety
interviewers experienced in systematic data collection
collected the data, having received a 3-day training
course in recruiting and interviewing, followed by a
4-day course at the WHO-CIDI training centre in Amsterdam. Extensive monitoring and quality checks took
place throughout the entire data collection period (Bijl
et al., 1998a).
N. Kaymaz et al. / Journal of Affective Disorders 98 (2007) 55–64
2.2. Psychosis ratings
Lifetime ratings from the 17 items of the CIDI core
psychosis sections, assessed at baseline, on delusions
(13 items) and hallucinations (4 items) were used (items
G1 –G13, G15, G16, G20, G21). These concern classic
psychotic symptoms involving, for example, persecution, thought interference, auditory hallucinations and
passivity phenomena. All these items can be rated in six
ways: “1”—no symptom, “2”—symptom present but
not clinically relevant (not bothered by it and not
seeking help for it), “3”—symptom result of ingestion
of drugs, “4”—symptom result of somatic disease,
“5”—true psychiatric symptom, “6”—symptom may
not really be a symptom because there appears to be
some plausible explanation for it. Because psychotic
symptoms are difficult to diagnose in a structured interview (Anthony et al., 1985; Cooper and Collacott, 1994;
Helzer et al., 1985), clinical reinterviews were conducted over the telephone by an experienced trainee
psychiatrist at the level of senior registrar for all individuals who had at least one rating of “5” or “6”, using
questions from the Structured Clinical Interview for
DSM-III-R (SCID), an instrument with proven reliability and validity in diagnosing schizophrenia (Spitzer
et al., 1992). CIDI ratings were corrected on the basis of
these clinical interviews, and the NEMESIS DSM-III-R
diagnoses of psychotic disorder are based on the data
from these clinical reinterviews; approximately 50%
of eligible people were reinterviewed. Of a possible
226 × 17 = 3842 CIDI ratings of psychotic symptoms in
the 226 individuals who were reinterviewed, changes
after clinical reinterview were introduced in 266 ratings
(6.9%).
Psychotic disorder was defined as any DSM-III-R
affective or non-affective psychotic diagnosis. Subclinical psychosis was defined as a CIDI rating of 5 on any
of the 17 CIDI core psychosis items.
2.3. Bipolar diagnoses and mania symptom scores
Lifetime manic symptoms were assessed using ratings
from the 11 items of the CIDI core mania section (F) at
baseline. All these items can be rated either “yes”(1) or
“no”(0). The manic symptom has to be present for at least
2 days. The mania section starts with assessing the presence of abnormally and persistently elevated, expansive
or irritable mood, which, if positive, results in further
questioning about other symptoms. Subclinical mania
was defined as at least one manic symptom thus assessed.
The diagnosis of bipolar disorder was defined as
DSM-III-R diagnosis on the basis of the CIDI. In order
to assess social impairment, item 6 of the Short Form
Health Survey, i.e. SF-36 (Van der Zee et al., 1993; Ware
et al., 1994) was used. The SF-36 measures the healthrelated quality of life and item 6 of the SF-36 inquires
about the interference of physical and emotional problems with normal social activities in the past 4 weeks (in
the analyses used as: none, a little, more than a little).
2.4. Predictors of comorbidity
Guided by previous research as described above, the
following predictors of psychotic symptoms occurring
in the context of subclinical mania were examined: age
(two groups around median split), sex, marital status
(living alone or not), education (4 levels), unemployment (present or absent), urbanisation (three levels),
family history of depression (present or absent), family
history of delusions and hallucinations (present or absent), lifetime use of cannabis (present or absent), lifetime use of other drugs (present or absent), childhood
trauma (present or absent), discrimination (present or
absent) and neuroticism (two groups around median
split of continuous score). Urbanisation was defined
following the standard classification of urbanisation
of place of residence according to the Dutch Central
Bureau of Statistics, based on the density of addresses
per km2 in an area (van Os et al., 2001). The three levels
of urbanicity represented respectively 0–499, 500–1499
and 1500 and more persons per km2.
Family history of delusions and hallucinations as
well as a family history of depression were assessed at
T1 by asking the subjects, for each first-degree relative,
whether any ever had presented with delusions or hallucinations (Van Os et al., 2003) or depression respectively, describing the symptoms associated with these
disorders. Lifetime use of cannabis and lifetime use of
other drugs (stimulant drug, cocaine, PCP and psychedelics) were assessed at baseline using the CIDI section
on alcohol and drugs (van Os et al., 2002). All these
items were classified dichotomously as either “no”(0) or
“yes”(1).
At baseline, subjects were asked using a semi-structured interview whether they had experienced any kind
of emotional, physical, psychological or sexual trauma
before age of 16 years. Subjects answered “yes” or “no”
to each of the questions and were able to give an indication about the frequency on a 6-point scale, “1” never,
“2” once, “3” sometimes, “4” regular, “5” often and “6”
very often. In the analyses, trauma was a priori dichotomised as follows; “no trauma” if the score on any item
≤ 3 and “trauma” if the score on any item N3 (Janssen
et al., 2004). Given a specific finding of early trauma
93
N. Kaymaz et al. / Journal of Affective Disorders 98 (2007) 55–64
predicting hallucinations in bipolar disorder (Hammersley et al., 2003), prediction was also tested using hallucinations only as the outcome. In order to assess
exposure to discrimination, subjects were asked if they
had experienced discrimination over the past year because of their skin colour or ethnicity, their sex, age,
appearance, disability or sexual orientation (Janssen
et al., 2003). Level of neuroticism was assessed with the
14-item Groningen Neuroticism Scale (Krabbendam
et al., 2002; Ormel, 1980; Ormel et al., 2001).
individuals. For the predictive analysis, the risk set
consisted of all individuals who had i) participated in the
baseline CIDI interview with the exception of the 218
individuals with a lifetime diagnosis of psychotic disorder or bipolar disorder and ii) who had had at least one
post-baseline CIDI interview (T1 or T2). After applying
these criteria, the risk set for the predictive analysis
consisted of 5501 individuals.
2.5. Data analyses
3.1. Sample description
All analyses were carried out using the software
package STATA, version 9 (StataCorp, 2005). The risk for
psychosis was calculated in individuals with and without
subclinical mania, and at different levels of the predictor
variables. In order to examine whether the risk for psychosis comorbidity was statistically different between
different levels of the predictor variables, a STATA risk
regression model of subclinical psychosis was used in
which the interaction between subclinical mania and the
predictor variable was estimated.
In order to assess whether the effect of any predictor
was additive to that of others, a variable representing the
sum score of significant predictor variables was constructed in order to assess whether comorbidity would
increase in a dose–response fashion with higher levels
of the predictor sum score.
In the predictive analyses, post-baseline onset of
bipolar disorder was the outcome variable and the presence or absence of one or more manic symptoms at
baseline (T0), outside any lifetime diagnosis of bipolar or
psychotic disorder, was the predictor variable. Predictive
power of the predictor variable was expressed as the
post-test probability (PPs) calculated with the DIAGTEST procedure in the STATA statistical program, and
calculated separately for i) those with subclinical psychosis at baseline and ii) those without subclinical psychosis at baseline. The PP is the likelihood of disease
(post-baseline bipolar disorder) given a positive test
result (baseline mania symptoms) (Sackett and Wennberg, 1997).
In the entire baseline sample of 7076 individuals, the
lifetime population prevalence of at least one manic and
one psychotic symptom were 4.1% and 4.2% respectively. The mean age of the risk set of 6857 individuals,
in which persons with any lifetime DSM-III-R bipolar
or psychotic disorder diagnosis (n = 218) had been
excluded, was 41.2 years (SD = 12.2) and 47% was male
(Table 1). In the risk set of 6857, the prevalence of
subclinical mania was 2.3% (mean symptom sum score:
0.1, SD = 0.65) and of subclinical psychosis was 2.8%.
2.6. Risk set
The risk set consisted of all individuals who had
participated in the baseline CIDI interview, with the
exception of the 218 individuals with a lifetime diagnosis of psychotic disorder (n = 107) or bipolar disorder
(n = 132, of whom 21 also had a diagnosis of psychotic
disorder) resulting in a sample for analysis of 6857
94
3. Results
Table 1
Risk set demographics, mania outcome and moderator variable
distributions
Variable
Frequencies (%)
or mean (SD)
Age
Male
Female
No partner
Unemployment
Education
Highest
High
Low
Lowest
At least one subclinical manic symptom
At least one subclinical psychotic symptom
Psychological problems interfere with functioning
A little
More than a little
Cannabis use
Use of other drugs
Neuroticism
Urbanicity
Lowest
High
Highest
Trauma
Discrimination
Family history of delusions/hallucinations
Family history depression
41.2 (12.2%)
3214 (47%)
3643 (53%)
1837 (27%)
451 (7%)
1865 (28%)
508 (8%)
2485 (37%)
1911 (28%)
156 (2.3%)
192 (2.8%)
16%
7%
629 (9%)
134 (2%)
3.8 (4.2%)
1164 (17%)
3071 (45%)
2622 (38%)
845 (12%)
1056 (15%)
279 (5%)
1438 (27%)
N. Kaymaz et al. / Journal of Affective Disorders 98 (2007) 55–64
Individuals with subclinical mania had a 17% risk of
subclinical psychosis, compared with only 2% in those
without (P b 0.000; Table 2). The subclinical psychosis
risk increased from 12%, 23% to 83% in individuals
with 1, 2 and 3 or more manic symptoms respectively
(test for trend: χ2 = 25.7, df = 1, P b 0.0001).
3.2. Predictors of comorbidity
Moderator
Age
In the risk regression model of subclinical psychosis,
there was a significant interaction between urbanicity and
subclinical mania, indicating a higher risk of comorbidity
between subclinical mania and subclinical psychosis for
subjects living in more urbanised areas (Table 3). Similar
positive interactions were apparent for cannabis use and a
family history of depression, indicating more comorbidity
in individuals using cannabis and in individuals with
familial loading for depression. Although there was no
significant interaction between childhood trauma and
subclinical psychosis, there was a positive interaction
using the outcome of hallucinations only, indicating a
higher risk for comorbidity between subclinical mania and
hallucinations in subjects with a history of recalled trauma
(Table 3). Although the risk for mania and psychosis
comorbidity was substantially higher in the expected direction for individuals who were younger (19% versus
14%), those using other drugs than cannabis (38% versus
15%), those with higher levels of neuroticism (20% versus
11%), those with a family history of psychosis (24%
versus 17%), those living alone (24% versus 13%) and
those in the highest level of social impairment due to
physical or psychological problems (29% versus 12% and
17%), none of these differences was statistically significant by conventional alpha (Table 3). No large or significant differences were seen for sex, unemployment, and
discrimination, whereas lower education was associated
with lower rather than higher comorbidity, although this
difference was not statistically significant (Table 3).
3.3. Additivity of predictors
A variable with score 0 to 3 representing the number
of predictors (cannabis, urbanicity highest level and
Table 2
Comorbidity of subclinical manic and subclinical psychotic symptoms
Subclinical mania
symptoms
Subclinical psychotic symptoms
Absent
Present
Test risk difference
6536
165 (2.5%)
129
27 (17.3%)
χ2 = 23.9, df = 1, P b 0.0001
Absent
Table 3
Comorbidity, stratified by moderator values
Lower median
group (18–40)
Higher median
group (41–64)
Sex
Women
Men
Caninabis
Absent
Present
Use other drugs
Absent
Present
Urbanicity#
Level 1
Level 2
Level 3
Neuroticism
Lower median
group (0–3)
Higher median
group (4–28)
Family history
Absent
psychosis^
Present
Family history
Absent
depressioin^
Present
Unemployment
Absent
Present
Living alone
Absent
Present
Childhood trauma
Absent
Present
Childhood trauma:
Absent
hallucinations only Present
Education
Highest
High
Low
Lowest
Discrimination
Absent
Present
Psychological
None
problems interfere Little
with social
More than
functioning
a little
Comorbidity
between
psychosis and
mania (%)*
Interaction
between the
moderator
and
psychosis@
χ2 ( p)
19
0.5 (0.47)
14
16
18
12
36
15
38
5
14
23
11
0.2 (0.63)
5.4 (0.002)
1.2 (0.28)
5.8 (0.018)
0.6 (0.44)
20
17
24
17
24
18
14
13
24
12
28
3
15
26
27
11
14
18
15
17
12
29
0.3 (0.60)
7.1 (0.0076)
0.5 (0.49)
1.9 (0.17)
2.6 (0.11)
4.1 (0.044)
2.9 (0.09)
0.9 (0.34)
0.34 (0.56)
*proportion of people with subclinical mania, dichotomously defined,
who also had evidence of subclinical psychosis, dichotomously
defined.
@ tests whether differences in risk are significantly different between
different levels of moderator variable.
# three levels representing number of persons per square kilometre:
1:0–499; 2:500–1499; 3:1500 and more.
^ family history of delusions and/or hallucinations (psychosis) or
depression, as reported by respondent.
Present
family history of depression) of each individual in the
risk set was constructed in order to assess the level of
comorbidity at each stratum. This revealed an increase
95
N. Kaymaz et al. / Journal of Affective Disorders 98 (2007) 55–64
Table 4
Prediction of incident bipolar disorder: the Post-test probabilities (PP)
of developing post-baseline DSM-III-R bipolar disorder in those with
subclinical mania at baseline with and without subclinical psychotic
symptoms
No subclinical
psychotic symptoms
With subclinical
psychotic symptoms
PP Post-baseline
bipolar disorder
95% confidence
interval
3.0%
2.5, 3.4
9.5%
4.7, 14.4
in subclinical psychosis comorbidity given the presence
of subclinical mania, increasing from 6% in those with
zero predictors to 12%, 30% and finally 63% in those
with 1, 2 and 3 predictors respectively.
3.4. Prediction of incident bipolar disorder
The PP of developing post-baseline DSM-III-R bipolar
disorder in those with subclinical mania at baseline given
the co-occurrence of subclinical psychosis was 9.5% (2 out
of 21 persons; 95%CI: 4.7, 14.4), considerably and significantly (as evidenced by non-overlapping 95% CIs) higher than the probability of developing bipolar disorder in the
group without subclinical psychosis at baseline (3 out of
101 persons; PP = 3.0%, 95% CI: 2.5, 3.4) (Table 4).
4. Discussion
This study examined the degree of comorbidity between subclinical mania and subclinical psychosis, the
predictors of subclinical comorbidity and prospective
post-baseline prediction of incident DSM-III-R diagnosis of bipolar disorder. Excluding the lifetime diagnosis
for clinical mania and psychosis at baseline, strong associations between subclinical mania and subclinical
psychosis were demonstrated. Subclinical comorbidity
was predicted by variables that previously have been
shown to increase the risk for psychosis such as urban
residence, cannabis use, familial loading for depression,
and victimisation (trauma with hallucinations as outcome). Subclinical comorbidity was not predicted by
other predictors, that previously have been shown to
increase the risk for psychosis such as age, gender, use of
other drugs, trait anxiety (neuroticism), family history of
psychosis, unemployment, single marital status, childhood trauma (without hallucinations), education, discrimination and social impairment due to physical and
psychological problems. Coexistence of subclinical
psychotic features with subclinical mania was much
more predictive of a future diagnosis of bipolar disorder.
96
4.1. Predictors of subclinical comorbidity
Some of our findings suggest that similar factors that
have previously been shown to increase the risk for
psychosis and mania at the clinical level, also increase
the risk for the co-occurrence for both symptoms at the
subclinical level. In our study there was a strong association between urban residence and subclinical comorbidity. Previous studies (Lewis et al., 1992; Marcelis et
al., 1998) have shown that urban residence is as a risk
factor for psychosis, whereas contradictory findings for
affective disorders exist. In some studies, urban residence has been found to be a risk factor for affective
disorders (Kessler et al., 1997; Sundquist et al., 2004),
but not in others (Mortensen et al., 2003). It was recently
shown that urban environment specifically impacts on
the psychotic, but not on the affective dimensions of
clinical bipolar disorder (Kaymaz et al., 2006), which is
compatible with the current findings in the subclinical
phenotype.
Use of cannabis has been shown to be a specific risk
factor for later psychosis (Andreasson et al., 1987;
Henquet et al., 2005; van Os et al., 2002; Zammit and
Lewis, 2004; Stefanis et al., 2004b; Fergusson et al.,
2003), as well as for the development of mania (Henquet
et al., 2006). In this study, only an association between
the use of cannabis and subclinical comorbidity was
found and not for the use of other drugs.
Although there are several studies showing common
genetic contributions to the variance in liability to psychosis and mania at a clinical level (Craddock et al.,
2006), in this study no association between familiar
loading for psychosis and comorbidity between psychosis and mania at a subclinical level was found. However,
a strong association between family history of depression and subclinical comorbidity was found. A significant association between trauma with hallucinations as
outcome and subclinical comorbidity was found, entirely
in line with the findings of a previous study (Hammersley et al., 2003). No association was found between
childhood trauma and comorbidity between mania and
psychosis at a subclinical level, whereas trauma has been
found to be a risk factor for psychosis (Read et al., 2005)
as well as bipolar disorder (Hammersley et al., 2003)
(Bebbington, 2004; Garno et al., 2005; Levitan et al.,
1998).
Although it is a robust epidemiological finding that
younger age and male sex increase the risk for psychosis
(Johns et al., 2004; Verdoux et al., 1998), inconsistent
findings exist for the impact of age and gender on the
prevalence in bipolar disorder. Some studies showed an
impact of age and gender on bipolar disorder (Gonzalez-
N. Kaymaz et al. / Journal of Affective Disorders 98 (2007) 55–64
Pinto et al., 2003; Yildiz and Sachs, 2003) whereas
others did not find any sex differences (Hendrick et al.,
2000). In this study, no association between age and
gender and subclinical comorbidity was found, nor was
there an association between neuroticism and subclinical comorbidity in spite of findings in literature of a
risk-increasing effect of neuroticism on psychosis
(Krabbendam et al., 2002; van Os et al., 2001) and
mania (Lozano and Johnson, 2001). No associations
were found between lower educational achievement,
single marital status, discrimination and unemployment
in this study, although previous studies have shown impact of these findings on psychosis and increased mental
disorders (Chakraborty and McKenzie, 2002; Janssen et
al., 2003; Karlsen and Nazroo, 2002; van Os et al., 2000a;
van Os et al., 1995). Although for the above mentioned
variables no significant association with subclinical comorbidity was found, the risk for mania and psychosis
comorbidity was nevertheless substantially higher in the
expected direction for individuals who were younger,
those using other drugs than cannabis versus those who
did not use, those with higher levels of neuroticism versus
those with a lower level, those with a family history of
psychosis versus those without a family history of psychosis, those living alone versus those living together and
those in the highest level of social impairment due to
physical or psychological problems versus none, little or
more than little. Therefore, the inconclusive findings
regarding risk of comorbidity may be likely due to insufficient statistical power. For example, the interaction with
family history of depression was significant, whereas the
interaction with family history of psychosis was not,
despite similar differences in comorbidity. The reason for
this was that a family history of psychosis was much rarer
than a family history of depression, yielding lower power.
There was a positive interaction between the variable
social impairment due to physical and psychological
problems and subclinical psychosis, indicating that for a
given level of subclinical mania, the coexistence of subclinical psychotic symptoms was more predictive of social impairment, but this interaction was not significant.
To summarise, some of the findings suggest that
genetic and non-genetic risk factors for psychosis and
mania overlap on lower levels of the continuum, mirroring the risk factors for the clinical disorders. These
shared aetiological factors may explain the substantial
comorbidity between both disorders.
4.2. Impact of subclinical comorbidity
In addition, comorbidity at a subclinical level was
predictive of a future diagnosis of bipolar disorder. The
probability of developing incident bipolar disorder was
3% in the absence of subclinical psychotic symptoms
and 10% in the presence of psychotic symptoms, indicating approximately a 7% higher risk of developing
future bipolar disorder.
The risk factors for comorbidity that were studied,
such as family history, use of cannabis, victimisation,
neuroticism and indicators of the wider social environment have all been described before as risk factors for
psychosis (Van Os et al., 2005), and thus may act in part
by facilitating the formation of more “toxic” combinations of subclinical symptoms of mania and psychosis in
the general population with higher probability of transition to clinical disorders. Thus, subclinical symptoms
can be seen as intermediary phenotypes of a mood continuum that after exposure to additional risk factors may
progress to a full-blown disorder (Hanssen et al., 2005),
specifically in the presence of psychotic symptoms at
subclinical level.
4.2.1. Methodological issues
This study mainly reports on cross-sectional associations between symptoms of mania and psychosis and
demographic and other risk variables and lacks information on temporal relationships between these variables.
However, an important finding was that subclinical psychosis was associated with a greater risk of transition to
bipolar disorder over time.
Psychosis includes also dimensions of negative symptoms and disorganisation in addition to positive symptoms. However, the CIDI does not include adequate
assessments of these dimensions so that their role could
not be investigated.
Another limitation of the used instrument is that during
this study, only respondents were interviewed and not their
relatives. Therefore, under-reporting of (hypo) manic
symptoms could have occurred, since these symptoms
often are experienced as ego-syntonic by the patient,
whereas family members may provide important contextual information. Another limitation of our study is that a
lifetime prevalence of a single manic symptom for 2 days,
as only assessed by lay interviews, may be underestimated; however there is no reason to assume that this
would have affected comorbidity rates and predictors
thereof.
Psychotic symptom ratings were assessed by lay
interviewers (CIDI ratings of 1–4), whereas the other
CIDI ratings (5–6) were assessed by clinicians through
telephone interviews in approximately 50% of eligible
cases. No clinical reinterview ratings were carried out
for the mania symptoms. It is possible that with lay
interviewer ratings a degree of misclassification did
97
N. Kaymaz et al. / Journal of Affective Disorders 98 (2007) 55–64
occur. However, the CIDI was designed for trained
interviewers who are not clinicians and has been found
to have high inter-rater reliability (Cottler, 1991;
Wittchen et al., 1991), and high test–retest reliability
(Semler et al., 1987; Wacker et al., 1990; Wittchen,
1994).
Bias could also have been introduced by differential
rating of psychosis due to incomplete clinical reinterview rates at baseline. However, excluding the individuals who were eligible for clinical reinterview but who
had not been contacted, thus leaving only those who had
been rated by clinicians, did not affect the results.
Our findings depend in part on the assumption that
the lay interviewers, using the CIDI, were enough experienced in the discrimination of manic and psychotic
symptoms. This assumption has face value since i) the
lay interviewers were trained during a three-day course,
ii) there was an extensive monitoring and quality checks
throughout the entire data collection period (Bijl et al.,
1998a), iii) individuals who screened positive for psychosis were reinterviewed by clinicians. The CIDI has
been designed for trained interviewers who are not
clinicians and has been found to have a high inter-rater
reliability (Cottler, 1991; Wittchen et al., 1991) and high
test–retest reliability (Semler et al., 1987; Wacker et al.,
1990; Wittchen, 1994). Nevertheless, a degree of misclassification may have occurred, although again it is
difficult to see how this would have affected the findings
relating to predictors of comorbidity.
Acknowledgements
Funded by the Dutch Department of Health.
References
Akiskal, H.S., Bourgeois, M.L., Angst, J., et al., 2000. Re-evaluating the
prevalence of and diagnostic composition within the broad clinical
spectrum of bipolar disorders. J. Affect. Disord. 59 (Suppl 1), S5–S30.
Alegria, M., Bijl, R.V., Lin, E., et al., 2000. Income differences in
persons seeking outpatient treatment for mental disorders: a
comparison of the United States with Ontario and The Netherlands.
Arch. Gen. Psychiatry 57, 383–391.
Andreasson, S., Allebeck, P., Engstrom, A., et al., 1987. Cannabis and
schizophrenia. A longitudinal study of Swedish conscripts. Lancet
2, 1483–1486.
Angst, J., 2002. Historical aspects of the dichotomy between manicdepressive disorders and schizophrenia. Schizophr. Res. 57, 5–13.
Anthony, J.C., Folstein, M., Romanoski, A.J., et al., 1985. Comparison
of the lay Diagnostic Interview Schedule and a standardized
psychiatric diagnosis. Experience in eastern Baltimore. Arch. Gen.
Psychiatry 42, 667–675.
Bailer, U., Leisch, F., Meszaros, K., et al., 2002. Genome scan for
susceptibility loci for schizophrenia and bipolar disorder. Biol.
Psychiatry 52, 40–52.
98
Bebbington, P., 2004. Recent findings in bipolar affective disorder.
Psychol. Med. 34, 767–776.
Bebbington, P., Wilkins, S., Jones, P., et al., 1993. Life events and
psychosis. Initial results from the Camberwell Collaborative
Psychosis Study. Br. J. Psychiatry 162, 72–79.
Bijl, R.V., Ravelli, A., van Zessen, G., 1998a. Prevalence of
psychiatric disorder in the general population: results of The
Netherlands Mental Health Survey and Incidence Study (NEMESIS). Soc. Psychiatry Psychiatr. Epidemiol. 33, 587–595.
Bijl, R.V., van Zessen, G., Ravelli, A., et al., 1998b. The Netherlands
Mental Health Survey and Incidence Study (NEMESIS): objectives
and design. Soc. Psychiatry Psychiatr. Epidemiol. 33, 581–586.
Bramon, E., Sham, P.C., 2001. The common genetic liability between
schizophrenia and bipolar disorder: a review. Curr. Psychiatry Rep.
3, 332–337.
Brown, A.S., van Os, J., Driessens, C., et al., 2000. Further evidence of
relation between prenatal famine and major affective disorder. Am.
J. Psychiatry 157, 190–195.
Cardno, A.G., Rijsdijk, F.V., Sham, P.C., et al., 2002. A twin study of
genetic relationships between psychotic symptoms. Am. J.
Psychiatry 159, 539–545.
Chakraborty, A., McKenzie, K., 2002. Does racial discrimination
cause mental illness? Br. J. Psychiatry 180, 475–477.
Chapman, L.J., Chapman, J.P., Kwapil, T.R., et al., 1994. Putatively
psychosis-prone subjects 10 years later. J. Abnorm. Psychology
103, 171–183.
Cooper, S.A., Collacott, R.A., 1994. Clinical features and diagnostic
criteria of depression in Down's syndrome. Br. J. Psychiatry 165,
399–403.
Coryell, W., Leon, A.C., Turvey, C., et al., 2001. The significance of
psychotic features in manic episodes: a report from the NIMH
collaborative study. J. Affect. Disord. 67, 79–88.
Cottler, L.B., 1991. The CIDI and CIDI-substance abuse module
(SAM): cross-cultural instruments for assessing DSM-III, DSMIII-R and ICD-10 criteria. NIDA Res. Monogr. 105, 220–226.
Craddock, N., O'Donovan, M.C., Owen, M.J., 2006. Genes for
schizophrenia and bipolar disorder? Implications for psychiatric
nosology. Schizophr. Bull. 32, 9–16.
Fergusson, D.M., Horwood, L.J., Swain-Campbell, N.R., 2003.
Cannabis dependence and psychotic symptoms in young people.
Psychol. Med. 33, 15–21.
Garno, J.L., Goldberg, J.F., Ramirez, P.M., et al., 2005. Impact of
childhood abuse on the clinical course of bipolar disorder. Br. J.
Psychiatry 186, 121–125.
Goldberg, D., Williams, P., 1988. User's Guide to the GHQ. NFERNelson, Windsor.
Gonzalez-Pinto, A., van Os, J., Perez de Heredia, J.L., et al., 2003.
Age-dependence of Schneiderian psychotic symptoms in bipolar
patients. Schizophr. Res. 61, 157–162.
Goodwin, R.D., Fergusson, D.M., Horwood, L.J., 2003. Neuroticism
in adolescence and psychotic symptoms in adulthood. Psychol.
Med. 33, 1089–1097.
Green, E.K., Raybould, R., Macgregor, S., et al., 2005. Operation of
the schizophrenia susceptibility gene, neuregulin 1, across
traditional diagnostic boundaries to increase risk for bipolar
disorder. Arch. Gen. Psychiatry 62, 642–648.
Hammersley, P., Dias, A., Todd, G., et al., 2003. Childhood trauma and
hallucinations in bipolar affective disorder: preliminary investigation. Br. J. Psychiatry 182, 543–547.
Hanssen, M., Bak, M., Bijl, R., et al., 2005. The incidence and outcome
of subclinical psychotic experiences in the general population. Br. J.
Clin. Psychol. 44, 181–191.
N. Kaymaz et al. / Journal of Affective Disorders 98 (2007) 55–64
Helzer, J.E., Robins, L.N., McEvoy, L.T., et al., 1985. A comparison of
clinical and diagnostic interview schedule diagnoses. Physician
reexamination of lay-interviewed cases in the general population.
Arch. Gen. Psychiatry 42, 657–666.
Hendrick, V., Altshuler, L.L., Gitlin, M.J., et al., 2000. Gender and
bipolar illness. J. Clin. Psychiatry 61, 393–396 (quiz 397).
Henquet, C., Krabbendam, L., Spauwen, J., et al., 2005. Prospective
cohort study of cannabis use, predisposition for psychosis, and
psychotic symptoms in young people. BMJ 330, 11.
Henquet, C., Krabbendam, L., de Graaf, R., et al., 2006. Cannabis use
and expression of mania in the general population. J. Affect.
Disord. (Jun 19; Electronic publication ahead of print).
Janssen, I., Hanssen, M., Bak, M., et al., 2003. Discrimination and
delusional ideation. Br. J. Psychiatry 182, 71–76.
Janssen, I., Krabbendam, L., Bak, M., et al., 2004. Childhood abuse as
a risk factor for psychotic experiences. Acta Psychiatr. Scand. 109,
38–45.
Johns, L.C., Cannon, M., Singleton, N., et al., 2004. Prevalence and
correlates of self-reported psychotic symptoms in the British
population. Br. J. Psychiatry 185, 298–305.
Johns, L.C., van Os, J., 2001. The continuity of psychotic experiences
in the general population. Clin. Psychol. Rev. 21, 1125–1141.
Jones, P., Rodgers, B., Murray, R., et al., 1994. Child development risk
factors for adult schizophrenia in the British 1946 birth cohort.
Lancet 344, 1398–1402.
Judd, L.L., Akiskal, H.S., Schettler, P.J., et al., 2003. The comparative
clinical phenotype and long term longitudinal episode course of
bipolar I and II: a clinical spectrum or distinct disorders? J. Affect.
Disord. 73, 19–32.
Karlsen, S., Nazroo, J.Y., 2002. Relation between racial discrimination, social class, and health among ethnic minority groups. Am. J.
Public Health 92, 624–631.
Kaymaz, N., Krabbendam, L., De Graaf, R., et al., 2006. Evidence that
the urban environment specifically impacts on the psychotic but
not the affective dimension of bipolar disorder. Soc. Psychiatry
Psychiatr. Epidemiol. (Jul 3; Electronic publication ahead of print).
Kendler, K.S., Gardner, C.O., 1997. The risk for psychiatric disorders
in relatives of schizophrenic and control probands: a comparison of
three independent studies. Psychol. Med. 27, 411–419.
Kendler, K.S., McGuire, M., Gruenberg, A.M., et al., 1993. The
Roscommon Family Study. II. The risk of nonschizophrenic nonaffective psychoses in relatives. Arch. Gen. Psychiatry 50, 645–652.
Kessler, R.C., Zhao, S., Blazer, D.G., et al., 1997. Prevalence,
correlates, and course of minor depression and major depression in
the National Comorbidity Survey. J. Affect. Disord. 45, 19–30.
Krabbendam, L., Janssen, I., Bak, M., et al., 2002. Neuroticism and low
self-esteem as risk factors for psychosis. Soc. Psychiatry Psychiatr.
Epidemiol. 37, 1–6.
Lenzenweger, M.F., 1994. Psychometric high-risk paradigm, perceptual aberrations, and schizotypy: an update. Schizophr. Bull. 20,
121–135.
Levitan, R.D., Parikh, S.V., Lesage, A.D., et al., 1998. Major
depression in individuals with a history of childhood physical or
sexual abuse: relationship to neurovegetative features, mania, and
gender. Am. J. Psychiatry 155, 1746–1752.
Lewis, G., David, A., Andreasson, S., et al., 1992. Schizophrenia and
city life. Lancet 340, 137–140.
Lozano, B.E., Johnson, S.L., 2001. Can personality traits predict
increases in manic and depressive symptoms? J. Affect. Disord. 63,
103–111.
Maier, W., Hofgen, B., Zobel, A., et al., 2005. Genetic models of
schizophrenia and bipolar disorder: overlapping inheritance or
discrete genotypes? Eur. Arch. Psychiatr. Clin. Neurosci. 255,
159–166.
Marcelis, M., Navarro-Mateu, F., Murray, R., et al., 1998. Urbanization and psychosis: a study of 1942–1978 birth cohorts in The
Netherlands. Psychol. Med. 28, 871–879.
Matsui, M., Sumiyoshi, T., Kato, K., et al., 2004. Neuropsychological
profile in patients with schizotypal personality disorder or
schizophrenia. Psychol. Rep. 94, 387–397.
Mortensen, P.B., Pedersen, C.B., Melbye, M., et al., 2003. Individual
and familial risk factors for bipolar affective disorders in Denmark.
Arch. Gen. Psychiatry 60, 1209–1215.
Myin-Germeys, I., Krabbendam, L., Jolles, J., et al., 2002. Are cognitive
impairments associated with sensitivity to stress in schizophrenia?
An experience sampling study. Am. J. Psychiatry 159, 443–449.
Neumann, C.S., Walker, E.F., 2003. Neuromotor functioning in
adolescents with schizotypal personality disorder: associations
with symptoms and neurocognition. Schizophr. Bull. 29, 285–298.
Ormel, J., 1980. Moeite met Leven of een Moeilijk Leven? University
of Groningen, Groningen.
Ormel, J., Oldehinkel, A.J., Brilman, E.I., 2001. The interplay and
etiological continuity of neuroticism, difficulties, and life events in
the etiology of major and subsyndromal, first and recurrent
depressive episodes in later life. Am. J. Psychiatry 158, 885–891.
Peters, E.R., Joseph, S.A., Garety, P.A., 1999. Measurement of
delusional ideation in the normal population: introducing the PDI
(Peters et al. Delusions Inventory). Schizophr. Bull. 25, 553–576.
Potash, J.B., Willour, V.L., Chiu, Y.F., et al., 2001. The familial
aggregation of psychotic symptoms in bipolar disorder pedigrees.
Am. J. Psychiatry 158, 1258–1264.
Poulton, R., Caspi, A., Moffitt, T.E., et al., 2000. Children's selfreported psychotic symptoms and adult schizophreniform disorder: a
15-year longitudinal study. Arch. Gen. Psychiatry 57, 1053–1058.
Read, J., van Os, J., Morrison, A.P., et al., 2005. Childhood trauma,
psychosis and schizophrenia: a literature review with theoretical
and clinical implications. Acta Psychiatr. Scand. 112, 330–350.
Robins, L.N., Wing, J., Wittchen, H.U., et al., 1988. The composite
international diagnostic interview. An epidemiologic instrument
suitable for use in conjunction with different diagnostic systems
and in different cultures. Arch. Gen. Psychiatry 45, 1069–1077.
Sackett, D.L., Wennberg, J.E., 1997. Choosing the best research design
for each question. BMJ 315, 1636.
Semler, G., Wittchen, H.U., Joschke, K., et al., 1987. Test–retest
reliability of a standardized psychiatric interview (DIS/CIDI). Eur.
Arch. Psychiatr. Neurol. Sci. 236, 214–222.
Smeets, R.M.W., Dingemans, P.M.A.J., 1993. Composite International
Diagnostic Interview (CIDI) Versie 1.1. World Health Organisation, Amsterdam.
Spitzer, R.L., Williams, J.B., Gibbon, M., et al., 1992. The Structured
Clinical Interview for DSM-III-R (SCID). I: history, rationale, and
description0. Arch. Gen. Psychiatry 49, 624–629.
StataCorp, 2005. STATA Statistical Software: Release 9.0. College
Station, Texas.
Stefanis, N.C., Hanssen, M., Smirnis, N.K., et al., 2002. Evidence that
three dimensions of psychosis have a distribution in the general
population. Psychol. Med. 32, 347–358.
Stefanis, N.C., Delespaul, P., Smyrnis, N., et al., 2004a. Is the excess
risk of psychosis-like experiences in urban areas attributable to
altered cognitive development? Soc. Psychiatry Psychiatr. Epidemiol. 39, 364–368.
Stefanis, N.C., Delespaul, P., Henquet, C., 2004b. Early adolescent
cannabis exposure and positive and negative dimensions of
psychosis. Addiction 99, 1333–1341.
99
N. Kaymaz et al. / Journal of Affective Disorders 98 (2007) 55–64
Sundquist, K., Frank, G., Sundquist, J., 2004. Urbanisation and
incidence of psychosis and depression: follow-up study of 4.4
million women and men in Sweden. Br. J. Psychiatry 184, 293–298.
Valles, V., Van Os, J., Guillamat, R., et al., 2000. Increased morbid risk
for schizophrenia in families of in-patients with bipolar illness.
Schizophr. Res. 42, 83–90.
Van der Zee, K., Sanderman, R., Heyink, J., 1993. De Psychometrische
kwaliteiten van de MOS 36-item Short Form Health Survey (SF36) in een Nederlandse populatie. Tijdschr. Soc. Gezondheidsz. 71,
183–191.
van Os, J., 1996. Risk factor assessment and prevention of psychosis.
Ned. Tijdschr. Geneeskd. 140, 1295–1299.
Van Os, J., 2000. Social influences on risk for disorder and natural
history. Curr. Opin. Psychiatry 13, 209–213.
Van Os, J., Jones, P.B., 2001. Neuroticism as a risk factor for
schizophrenia. Psychol. Med. 31, 1129–1134.
van Os, J., Takei, N., Castle, D.J., et al., 1995. Premorbid abnormalities
in mania, schizomania, acute schizophrenia and chronic schizophrenia. Soc. Psychiatry Psychiatr. Epidemiol. 30, 274–278.
Van Os, J., Jones, P., Sham, P., et al., 1998. Risk factors for onset and
persistence of psychosis. Soc. Psychiatry Psychiatr. Epidemiol. 33,
596–605.
van Os, J., Driessen, G., Gunther, N., et al., 2000a. Neighbourhood
variation in incidence of schizophrenia. Evidence for person–
environment interaction. Br. J. Psychiatry 176, 243–248.
Van Os, J., Hanssen, M., Bijl, R.-V., et al., 2000b. Straus (1969)
revisited: a psychosis continuum in the general population?
Schizophr. Res. 45, 11–20.
van Os, J., Hanssen, M., Bijl, R.V., et al., 2001. Prevalence of
psychotic disorder and community level of psychotic symptoms:
an urban-rural comparison. Arch. Gen. Psychiatry 58, 663–668.
van Os, J., Bak, M., Hanssen, M., et al., 2002. Cannabis use and
psychosis: a longitudinal population-based study. Am. J. Epidemiol.
156, 319–327.
Van Os, J., Hanssen, M., Bak, M., et al., 2003. Do urbanicity and
familial liability coparticipate in causing psychosis? Am. J.
Psychiatry 160, 477–482.
100
Van Os, J., Krabbendam, L., Myin-Germeys, I., et al., 2005. The
Schizophrenia Envirome. Curr. Opin. Psychiatry 18, 141–145.
Verdoux, H., van Os, J., Maurice-Tison, S., et al., 1998. Is early
adulthood a critical developmental stage for psychosis proneness?
A survey of delusional ideation in normal subjects. Schizophr. Res.
29, 247–254.
Voglmaier, M.M., Seidman, L.J., Niznikiewicz, M.A., et al., 2000.
Verbal and nonverbal neuropsychological test performance in
subjects with schizotypal personality disorder. Am. J. Psychiatry
157, 787–793.
Wacker, H.R., Battegay, R., Mullejans, R., et al., 1990. In: Stefanis, C.N.,
Rabavilas, A.D., Soldatos, C.R. (Eds.), Using the CIDI in the general
population. In Psychiatry: a world perspective. Elsevier Science
Publishers, Amsterdam, pp. 138–143.
Walss-Bass, C., Escamilla, M.A., Raventos, H., et al., 2005. Evidence
of genetic overlap of schizophrenia and bipolar disorder: linkage
disequilibrium analysis of chromosome 18 in the Costa Rican
population. Am. J. Med. Genet., B Neuropsychiatr. Genet.
Ware, J.E., Gandek, B., Group, I.P., 1994. The SF-36 Health Survey:
development and use in mental health research and the IQOLA
project. Int. J. Ment. Health 23, 49–73.
Wittchen, H.U., 1994. Reliability and validity studies of the WHOComposite International Diagnostic Interview (CIDI): a critical
review. J. Psychiatr. Res. 28, 57–84.
Wittchen, H.U., Robins, L.N., Cottler, L.B., et al., 1991. Cross-cultural
feasibility, reliability and sources of variance of the Composite
International Diagnostic Interview (CIDI). The Multicentre WHO/
ADAMHA Field Trials. Br. J. Psychiatry 159, 645–653 (658).
Yildiz, A., Sachs, G.S., 2003. Age onset of psychotic versus nonpsychotic bipolar illness in men and in women. J. Affect. Disord.
74, 197–201.
Zammit, S., Lewis, G., 2004. Exploring the relationship between
cannabis use and psychosis. Addiction 99, 1353–1355.
Kaymaz et al.
Evidence That Patients With Single Versus
Recurrent Depressive Episodes Are Differentially
Sensitive to Treatment Discontinuation:
A Meta-Analysis of Placebo-Controlled Randomized Trials
Nil Kaymaz, M.D.; Jim van Os, M.D., Ph.D.;
Anton J. M. Loonen, M.D., Pharm.D., Ph.D.; and Willem A. Nolen, M.D., Ph.D.
Background: Antidepressants are effective in
the prevention of relapse after remission from an
acute depressive episode. It is unclear, however,
to what degree duration of the continuation phase,
level of abruptness of antidepressant discontinuation, or the number of previous episodes moderate the prophylactic effect of antidepressants.
Data Sources: Searches were conducted
to identify all published randomized, placebocontrolled, double-blind clinical trials available
for review by May 2007 on the efficacy of continuation or maintenance treatment of major depressive disorder with either selective serotonin
reuptake inhibitors (SSRIs) or tricyclic antidepressants (TCAs) that included patients entering a
maintenance phase after achieving remission from
the acute phase. The MEDLINE and EMBASE
databases were searched using the terms depression, antidepressants, discontinuation, and maintenance treatment; this was followed by reference
checks of articles thus identified. In addition, the
Cochrane Library was also searched using the
same terms. Some authors of the identified
papers were contacted for specific data.
Data Synthesis: Data were collected from 30
trials with 4890 participating patients. The overall
reduction of relapse risk in the maintenance phase
was highly significant for both SSRIs (OR = 0.24,
95% CI = 0.20 to 0.29) and TCAs (OR = 0.29,
95% CI = 0.23 to 0.38) over 1 year of follow-up
of maintenance treatment. The prophylactic effect
appeared to be constant over the length of the
continuation phase. Recurrent episode patients
experienced less protection from antidepressants
over the maintenance phase (OR = 0.37, 95%
CI = 0.31 to 0.44) than single episode patients
(OR = 0.12, 95% CI = 0.06 to 0.26).
Conclusions: Antidepressants robustly reduce
relapse risk in the maintenance phase, regardless
of a number of clinical and pharmacologic factors. There is evidence, however, that with increasing number of episodes, patients develop
a relative resistance against the prophylactic
properties of antidepressant medication.
(J Clin Psychiatry 2008;69:1423–1436)
Received Nov. 28, 2006; accepted Dec. 16, 2007. From the
Department of Psychiatry, University of Maastricht, Maastricht, the
Netherlands (Drs. Kaymaz and van Os); the Department of Pharmacy,
Division of Pharmacotherapy and Pharmaceutical Care (Dr. Loonen)
and the Department of Psychiatry, University Medical Center Groningen
(Dr. Nolen), University of Groningen, Groningen, the Netherlands; and
the Division of Psychological Medicine, Institute of Psychiatry, London,
United Kingdom (Dr. van Os).
The authors thank the anonymous referees for their detailed and
extremely helpful comments on earlier versions of the manuscript.
Dr. van Os has received grants from or is a speaker for Eli Lilly,
Bristol-Myers Squibb, Lundbeck, Organon, Janssen-Cilag,
GlaxoSmithKline, Otsuka, and AstraZeneca. Dr. Loonen is a speaker or
advisory board member for Eli Lilly, Bristol-Myers Squibb, Lundbeck,
Janssen-Cilag, Wyeth, Bional, and AstraZeneca. Dr. Nolen has received
grants from the Netherlands Organization for Health Research and
Development, Netherlands Organization for Scientific Research, Stanley
Medical Research Institute, AstraZeneca, Eli Lilly, GlaxoSmithKline,
and Wyeth and is a speaker or advisory board member for AstraZeneca,
Cyberonics, Eli Lilly, GlaxoSmithKline, Johnson & Johnson, Pfizer, and
Servier. Dr. Kaymaz reports no financial or other relationship relevant to
the subject of this article.
Corresponding author and reprints: Prof. Willem A. Nolen, M.D.,
Ph.D., Department of Psychiatry 5.17, University Medical Center
Groningen, P.O. Box 30.0001, 9700 RB Groningen, the Netherlands
(e-mail: [email protected]).
D
epression often manifests itself as a chronic or a
recurrent illness; 15% to 20% of depressed patients experience a chronic course, and 75% to 80% of patients experience recurrent episodes.1,2 Therefore, treatment should focus not only on improving symptoms of
the acute episode but also on the prevention of relapse (return of symptoms of the index episode) and recurrence
(the development of a subsequent episode).3 The efficacy
of antidepressants in treating acute episodes (4–6 and
up to 12 weeks) has been well established in placebocontrolled studies, although the effect sizes for antidepressant treatment are only moderately larger than for placebo.4,5 Moreover, antidepressants are effective in longterm treatment. In a 2003 meta-analysis of 31 randomized
trials (4410 participants), Geddes et al.6 showed that continuing antidepressant therapy consistently reduced the
risk of relapse and recurrence by 70% compared with continuation with placebo, and this seemed to be similar for
all classes of antidepressants. They also found no differences in relapse and recurrence rates between patients
101
Antidepressant Discontinuation and Recurrent Episodes
with shorter (1–2 months) and longer (4–6 months)
treatment after having achieved remission and prior to
randomization or between patients with relatively short (6
months) versus longer (up to 36 months) follow-up, suggesting that the reduced risk is largely independent of the
duration of treatment before randomization and the duration of the randomly allocated therapy.
Besides the question of how long treatment should be
continued once remission has been attained, another important question facing clinicians in their daily practice
is, first, to what degree do patients with multiple episodes
acquire additional vulnerabilities that could make them
more vulnerable not only to discontinuation of the antidepressant per se, but also to the mode of discontinuation
(gradual or acute)? The rationale for such a distinction between single and multiple episode patients was proposed
by Post and colleagues7 in an attempt to account for several phenomena observed in the course of affective illness; they emphasized the importance of preventing episodes with prophylactic treatment to inhibit sensitization.
According to the sensitization model, a subgroup of patients exists who with each recurrent episode becomes
more vulnerable, or “sensitized,” to affective episode precipitants. The authors suggested that these patients show
characteristics of a sensitization or kindling-like process,
in which the biochemical and physiologic processes involved in the illness become progressively more easily
triggered by the same circumstances or precipitants compared to the first episode.
Therefore, in the literature on relapse prevention by
antidepressants, several questions remain unanswered. It
is unclear (1) how long treatment should be continued after remission is achieved (e.g., 6 months, 1 year, longer?),
(2) whether treatment should be continued longer in patients with multiple episodes compared to single episodes,
and (3) how the antidepressant should be discontinued:
can it be done abruptly or within 1 week, or should it be
tapered off gradually for 1 or more weeks? In the present
study, these questions were addressed, with the hypothesis that multiple episode patients would be more sensitive to antidepressant discontinuation, in particular abrupt
discontinuation.
METHOD
In 1988, the MacArthur Foundation Research Network
on the Psychobiology of Depression convened a task
force to examine the ways in which change points in the
course of depressive illness until that time had been described and the extent to which inconsistencies in these
descriptions might hinder research on this disorder.8 Consistent conceptualization and empirical validation of these
terms were considered desirable for the following reasons: (1) to be able to improve design, interpretation, and
comparison of studies on natural course and treatment;
102
(2) to be able to clarify the relationship between biological
and psychological correlates of illness; (3) to create improved guidelines for evaluation of clinical efficacy of
drugs and other treatments by regulatory agencies; (4) to
be able to conduct empirically based revisions of diagnostic criteria; and (5) to be able to develop improved
treatment guidelines for clinical practice. Guided by these
statements, the following definitions were constructed.
Definitions
In the literature, the terms response, remission, relapse,
and recurrence and the terms describing the different
treatment phases (acute treatment, continuation treatment,
and maintenance treatment) are not uniformly defined. In
this article, the definitions by Frank and colleagues8 were
followed. The first phase in treatment is acute treatment,
aimed at the suppression of the depressive symptoms. Response is defined as a clinically significant reduction of
symptoms of depression (e.g., at least 50% reduction of
the score on the Hamilton Rating Scale for Depression
[HAM-D]), and remission is defined as the remaining of
no or only minimal symptoms of depression (e.g., a score
of less than 9 points on the HAM-D). During the first 6
months after remission is achieved, the underlying illness
may still be present and discontinuation of the treatment
can cause reappearance of symptoms of the original episode; this is called a relapse. In this context, continuation
treatment is the term applied to the prevention of a relapse, i.e., the treatment phase during the first 6 months
after having achieved remission. After these 6 months of
remission, the underlying disorder is considered resolved,
and when a new depressive episode appears after this period, it is called recurrence. The term maintenance treatment is applied for a treatment with antidepressants that
aims at the prevention of a recurrence.
Types of Studies
We wished to identify all published randomized,
placebo-controlled, double-blind clinical trials available
for review by May 2007 on the efficacy of continuation or
maintenance treatment of major depressive disorder with
either selective serotonin reuptake inhibitors (SSRIs) or
tricyclic antidepressants (TCAs) (the 2 groups of comparable and widely used antidepressants with the largest
number of trials) in patients who had achieved remission
during acute treatment with these antidepressants.
Search Strategy for Identification of Studies
A MEDLINE and EMBASE computerized search
using the terms depression, antidepressants, discontinuation, and maintenance treatment was conducted and was
supplemented with references cited in reports so identified
(reference checking). In addition, the Cochrane Library
was also searched using the same terms. Some authors of
the identified papers were contacted for specific data.
Kaymaz et al.
Types of Participants
Trials were eligible for the review if they included patients with major depressive disorder, single or recurrent
episode, who were treated with an antidepressant (SSRI
or TCA) until remission was attained. These patients had
then to be followed up during the continuation and subsequent maintenance phases.
Types of Interventions
Patients obtaining remission and entering the continuation or the maintenance phase had to be randomly divided into at least 2 groups, with at least 1 group receiving
active treatment with an antidepressant (SSRI or TCA)
and another group receiving placebo, under double-blind
conditions.
Data Selection
The search results and the data extracted from the trial
reports on participant characteristics, intervention details,
and outcome measures were checked by 2 reviewers (N.K.
and A.J.M.L.) before analyses. When specific data were
not given in the report, the authors were contacted; some
authors responded to the 2 efforts we made to contact
them, and some did not.
Data were entered into the STATA, version 9, software
program (StataCorp, 2005)9 for further analyses.
Methodological Quality of Included Studies
The assessment of the methodological quality of the
trials was carried out by 2 reviewers (N.K. and A.J.M.L.)
who used the same checklist. The most crucial aspects
of the included trials for the internal validity were (1) the
method of randomization and achievement of a doubleblind condition and (2) the reporting of withdrawals and
dropout rates and use of a suitable survival analysis.
Data Aggregation
The following data were extracted and tabulated from
each paper (see Table 1): number of previous episodes,
duration of active treatment, duration of stabilization/
continuation treatment, number of subjects, duration of
follow-up, and duration of withdrawal of medication. Besides these tabulated data, the numbers of individuals
at risk over each follow-up interval (at 3, 6, 9, and 12
months) were extracted from the publications whenever
they were mentioned or, if they were not, from the survival
curves in these publications.
Data Analysis
Data from the selected randomized controlled trials
were combined to estimate the pooled odds ratio (OR)
with 95% confidence intervals (CIs) using a randomeffects model. The presence of heterogeneity across trials
was evaluated using a χ2 test for homogeneity, testing
whether individual trials results varied more than could be
explained by chance alone. When significant heterogeneity was found, possible causes were explored. In addition, meta-regression was used to assess the degree to
which important clinical and study quality factors impacted on the meta-analytic results. Meta-regression extends a random-effects meta-analysis to estimate the extent to which 1 or more covariates, with values defined for
each study in the analysis, explain heterogeneity in the
treatment effects. In meta-regression, the log odds ratio,
representing the antidepressant effect size, is regressed on
the variables that are hypothesized to explain heterogeneity in the treatment effects.
Potential publication bias was tested for using Begg’s
test for asymmetry.
As to the question of effectiveness of continuation
and maintenance treatment with antidepressants, relapse/
recurrence rates after randomization to either continuation
of the antidepressant or switch to placebo were compared,
whenever possible, over a 1-year follow-up period separately at 3, 6, 9, and 12 months. If relapse/recurrence rates
at these assessment points were not presented in the results
of the publication, they were calculated when possible,
e.g., by estimating these rates from the survival curves that
were presented as figures in most of the publications. The
effect of time at follow-up on relapse/recurrence rates was
examined using meta-regression analysis with time at
follow-up (3, 6, 9, or 12 months) as independent variable
and the log odds ratio representing antidepressant effect
size as the dependent variable. In order to compile the
dataset for this analysis, each study with data (i.e., number
of relapses/recurrences) at 3, 6, 9, and 12 months was
separated into 4 different studies of 3 months’ duration
each with the variable “time” denoting whether follow-up
was at 3, 6, 9, or 12 months. This way, the effect of time on
relapse rate could be assessed in the meta-regression of the
effect sizes, using the STATA meta-regression routine in
STATA (StataCorp, 2005, version 9).9 Using this dataset,
controlling for time, meta-regression was also applied to
investigate whether duration of continuation treatment, after achievement of remission, was associated with the risk
of relapse/recurrence after discontinuation of antidepressants. To this end, studies were subdivided according to
the duration of continuation treatment prior to randomization as follows: less than 1 month, 1 to 3 months, > 3 to 6
months, and > 6 months (i.e., beyond the continuation
phase of 6 months recommended by the guidelines).
Similarly, meta-regression, controlling for time, was
used to examine whether mode of discontinuation in controlled studies affected relapse rates by comparing abrupt
discontinuation (< 1 week) with gradual discontinuation
(tapering for ≥ 1 week).
Finally, meta-regression, controlling for time, was used
to assess whether clinical features such as the number of
previous episodes moderated relapse rates. To this end, the
studies were subdivided further into those involving
103
Antidepressant Discontinuation and Recurrent Episodes
patients with no previous episode of depression versus
those involving patients with multiple episodes, i.e., at
least 1 previous episode prior to the index episode. As we
had hypothesized that the patients most sensitive to the
antidepressant discontinuation would be those with the
combination of multiple episodes and an abrupt mode of
discontinuation, a number of previous episodes–by–mode
of discontinuation interaction term was fitted in the metaregression model.
In the event that a variable in the meta-regression was
found to impact significantly on the meta-analytic result,
separate meta-analyses were carried out for the different
strata of this variable, so as to clarify the direction and
magnitude of group differences in effect size.
A test for selection bias was carried out, visualizing the
included studies in a Begg’s funnel plot with pseudo 95%
confidence limits.
RESULTS
Identified Studies
The search process yielded 44 studies possibly satisfying the inclusion criteria, i.e., the continuation/
maintenance treatment was with either an SSRI or a TCA.
Fourteen studies were excluded due to methodological
limitations: 1 because the study was single blind,10 4 because there was no placebo control group,11–14 1 because
the antidepressant in the acute or continuation phase was
not the same antidepressant as in the maintenance phase,15
4 because lithium was added during the acute phase to
achieve remission,16–19 1 because the study was not formally published,20 and 1 because patients with bipolar disorder were also included.21 We also excluded 1 study22 of
maintenance treatment with a 5-year follow-up as it included patients who were also part of the 3-year maintenance study by Frank et al.45 One study (McGrath et al.23)
was excluded as it did not mention the number of relapses
separately for the medication and placebo arms in this
study. Only the active treatment arms, without crossover,
and the placebo arm of the studies with different assignment groups in which relapse rates are mentioned separately for the different arms of the study were included.
The characteristics of the 30 studies included in this
meta-analysis are listed in Table 1.24–53 Some of the studies were 4-armed or 5-armed studies, in which medication
was compared to psychotherapy or their combination;
from these studies, we only included data from the medication arm and the placebo arm. The 30 trials involved a
total of 4890 patients, performed in both primary and secondary care settings, and described 2749 patients who
continued with the antidepressant and 2141 patients who
were switched to placebo. Fifteen studies were found in
which the allocated drug was an SSRI, with a total of
2984 patients, and 15 studies were found in which the allocated drug was a TCA, with a total of 1906 patients.
104
There were no 3-arm studies with 2 medication arms
(e.g., both an SSRI and a TCA) versus placebo.
Duration of Continuation Phase
In order to examine the impact of the prerandomization period (e.g., the duration of the continuation
phase), studies were divided in 4 subgroups with a duration of less than 1 month, 1 to 3 months, > 3 to 6 months
of continuation phase, and > 6 months of continuation
plus maintenance phase.
In the subgroup of trials with a continuation treatment
of less than 1 month (i.e., remission achieved for less than
1 month) prior to randomization, 9 trials (1261 patients in
total) were found, and of these, 7 trials (954 patients) provided information at 3 months of follow-up, 6 trials (727
patients) provided information at 6 months of follow-up,
3 trials (339 patients) provided information at 9 months
of follow-up, and 2 trials (101 patients) provided information at 12 months of follow-up. In the subgroup of trials with 1 to 3 months of continuation treatment prior to
randomization, 2 trials (95 patients) provided information
at 3 (85 patients), 6 (70 patients), 9 (63 patients), and 12
(59 patients) months of follow-up. In the subgroup of trials with > 3 to 6 months of continuation treatment prior
to randomization, 17 trials (3194 patients in total) were
found, and of these 17 trials, 14 (2384 patients) provided
information at 3 months of follow-up, 13 trials (2104 patients) at 6 months of follow-up, 11 trials (1846 patients)
at 9 months of follow-up, and 9 trials (1547 patients) at
12 months of follow-up. Only 1 study was found (Bialos
et al.38) with a clearly mentioned treatment period prior
to randomization of more than 6 months. Another study,
by Cook et al.,42 had a duration of 52 weeks, but did not
clarify how long the duration of the acute and continuation treatment lasted. These publications were included
in the subgroup of studies with a prerandomization period
of 3 to 6 months. To examine the impact of time at followup (3, 6, 9, and 12 months) on relapse/recurrence rates,
we have used meta-regression analysis on pooled results
of 30 trials in total (4890 patients) (see Figure 2), 23 trials
(3441 patients) providing information at 3 months of
follow-up (see Figure 3), 21 trials (2914 patients) at 6
months of follow-up (see Figure 4), 16 trials (2259
patients) at 9 months of follow-up (see Figure 5), and 8
trials (1710 patients) at 12 months of follow-up (see
Figure 6).
Number of Previous Depressive Episodes
With regard to the number of episodes prior to the
index period, 4 trials (301 participants) included patients
with no prior episodes, and 17 trials (2939 participants)
included patients with at least 1 episode prior to the index
episode. The remaining 9 studies of the total of 30 were
left out of the analysis, because some studies included
both single- and recurrent episode patients. Some studies
Kaymaz et al.
Table 1. Characteristics of the Included Studies
No. of Previous Episodes
≥2
Not reported
Not reported
≥2
≥2
Mixed
Duration of Active
Treatment Phase
(wk)
6
8
6
8
8
12
DSM-III-R
Not reported
12–14
MDD
MDD
DSM-III-R
DSM-IV
≥2
≥2
6
6–9
Klysner et al, 200233
Mindham et al, 197234
Klerman et al, 197435
Coppen et al, 197836
MDD
Depression
Depression
Depression
DSM-IV
MRC
DSM-II
MRC
0
Not reported
0
11 patients = 0; 21 patients ≥ 1
8
3–10
4–6
Not reported
Stein et al, 198037
Bialos et al, 198238
Kane et al, 198239
Glen et al, 198440
Prien et al, 198441
Cook et al, 198642
Georgotas et al, 198943
Rouillon et al, 198944
Frank et al, 199045
OADIG, 199346
Reynolds et al, 199947
Alexopoulos et al, 200048
Wilson et al, 200349
Gilaberte et al, 200150
Schmidt et al, 200051
MDD
Depression
Depression
Depression
Depression
MDD
MDD
MDD
Depression
MDD
MDD
MDD
MDD
MDD
MDD
Feighnerb
RDC
RDC
MRC
RDC
RDC
RDC
DSM-III
RDC
RDC
RDC
RDC, DSM-IV
DSM-III-R, HAM-D
DSM-III-R
DSM-IV
≥1
Chronic depression
≥2
0
≥1
≥1
≥3
≥1
≥2
50% = 1; 50% ≥ 1
Average 5
Most = 1 or > 2
0
≥1
Mixed
6
Not reported
Not reported
Average 81/2
Not reported
Not reported
7–9
8
Not reported
16
Not reported
Not reported
8
8
13
McGrath et al, 200652
MDD
DSM-IV
Chronic depression
12
Perahia et al, 200653
MDD
DSM-IV
≥1
12
Study
Montgomery et al, 198824
Doogan and Caillard, 199225
Montgomery and Rasmussen, 199226
Montgomery and Dunbar, 199327
Robert and Montgomery, 199528
Keller et al, 199829
Diagnosis
MDD
MDD
MDD
MDD
MDD
Chronic depression,
MDD, double depressiona
Diagnostic Criteria
DSM-III
DSM-III-R
DSM-III-R
DSM-III-R
DSM-III-R
DSM-III-R, HAM-D
Reimherr et al, 199830
MDD
Terra and Montgomery, 1998
Hochstrasser et al, 200132
31
a
Major depression with antecedent dysthymic disorder.
Feighner et al. 1972 criteria.
Abbreviations: CGI = Clinical Global Impressions scale, CGI-S = Clinical Global Impressions-Severity of Illness scale,
GAS = Global Assessment Scale, HAM-D = Hamilton Rating Scale for Depression, MADRS = Montgomery-Asberg Depression Rating Scale,
b
included only patients with a chronic depressive disorder
or did not specify the number of previous episodes.
Mode of Discontinuation of the Antidepressant
Antidepressant medication was discontinued abruptly
(< 1 week) in 22 trials (4320 participants) and gradually
(≥ 1 week) in 8 trials (630 participants). No studies
were found that included both abrupt and gradual
discontinuation.
Publication Bias
The distribution of effect size–related measures relative to sample size–related measures did not suggest publication bias in the studies used for this meta-analysis
(Figure 1).
Treatment Effects
Overall, the results showed that continuing antidepressant therapy consistently reduced the risk of relapse
(OR = 0.30, 95% CI = 0.25 to 0.35, p < .001) compared
to placebo (Figure 2), with no significant heterogeneity
detected (χ2 = 40.21, df = 29, p = .081). This effect was
highly significant for SSRIs (OR = 0.24, 95% CI = 0.20
to 0.29, p < .001) as well as for TCAs (OR = 0.29, 95%
CI = 0.23 to 0.38, p < .001).
Using meta-regression, the overall relapse-reducing
effects of the SSRIs were not significantly different from
those of the TCAs (meta-regression coefficient = –0.30,
95% CI = –0.78 to 0.17, p = .209). However, it should be
noted that none of the studies involved (next to placebo) a
head-to-head comparison of TCAs versus SSRIs.
105
Antidepressant Discontinuation and Recurrent Episodes
Duration of Stabilization/
Continuation Phase
(prerandomization phase)
(wk)
24
0
0
0
0
16
Duration of
Withdrawal of
Criteria for Relapse/Recurrence
Follow-Up (mo) Medication (wk)
HAM-D > 18
12
<1
Clinical criteria, CGI ≥ 4
11
3
MADRS ≥ 22
6
<1
CGI ≥ 4, clinical criteria
12
<1
MADRS ≥ 25, clinical criteria
6
<1
DSM-III criteria for at least 3 wk,
19
Maximum 3
CGI ≥ 4, HAM-D ≥ 4 points higher
than maintenance phase baseline
0
Fluoxetine = 102, placebo = 96
HAM-D ≥ 14 for 3 successive wk,
121/2
<1
DSM criteria for minimum 2 wk
18
Fluvoxamine = 109, placebo = 94 Clinical criteria
12
<1
16
Citalopram = 132, placebo = 132
MADRS ≥ 22 confirmed at 3–7 d,
12–19
<1
CGI ≥ 5
16
Citalopram = 60, placebo = 61
MADRS ≥ 22
12
<1
0
Amitriptyline = 50, placebo = 42
Clinical criteria on monthly screenings
6
<1
0
Amitriptyline = 25, placebo = 25
Clinical criteria
8
<1
6
Amitriptyline = 16, placebo = 16
Increase in symptoms sufficient
12
<1
to warrant hospital admission
2
Amitriptyline = 28, placebo = 27
Clinical criteria
6
<1
192
Amitriptyline = 7, placebo = 10
Clinical criteria
6
3
24
Imipramine = 6, placebo = 6
RDC, clinical criteria
24
<1
0
Amitriptyline = 8, placebo = 9
Clinical criteria
36
2
Not reported
Imipramine = 39, placebo = 34
RDC, GAS ≤ 60, clinical criteria
24
<1
≥ 52
TCA = 6, placebo = 9
Clinical criteria
8
4–8
16
Nortriptyline = 13, placebo = 23
RDC, HAM-D ≥ 16
12
<1
24
Maprotiline = 767, placebo = 374 MADRS > 27 or > 25 for 2 wk
12
<1
17
Imipramine = 28, placebo = 23
RDC, HAM-D ≥ 15, Raskin ≥ 7
36
<1
8
Dothiepin = 33, placebo = 36
MADRS >10, clinical criteria
24
<1
16
Nortriptyline = 28, placebo = 29
RDC
36
6
16
Nortriptyline = 22, placebo = 21
RDC, DSM-IV criteria, HAM-D ≥ 17
24
10
16–20
Sertraline = 56, placebo = 57
HAM-D ≥ 13, DSM-III-R clinical criteria
24
<1
24
Fluoxetine = 70, placebo = 70
HAM-D ≥ 18, CGI ≥ 4, DSM-III-R
12
<1
Not reported
Fluoxetine = 189, placebo = 122
Criteria for depressive episode determined
25
<1
by the SCID-P, CGI-S ≥ 2
24
Fluoxetine = 131, placebo = 131
2 wk of ratings of less than
8
<1
“much improved” on the CGI,
compared with ratings at study entry
1
24
Duloxetine = 136, placebo = 142
CGI ≥ 2 compared to baseline,
/2
1
MINI criteria for MDD
Abbreviations continued: MDD = major depressive disorder, MINI = Mini-International Neuropsychiatric Interview, MRC = U.K. Medical
Research Council, OADIG = Old Age Depression Interest Group, Raskin = Raskin Depression Scale, RDC = Research Diagnostic Criteria,
SCID-P = Structured Clinical Interview for DSM-IV patient version, TCA = tricyclic antidepressant.
No. of Patients in Trial
Fluoxetine = 88, placebo = 94
Sertraline = 185, placebo = 110
Citalopram = 105, placebo = 42
Paroxetine = 68, placebo = 67
Citalopram = 152, placebo = 74
Sertraline = 77, placebo = 84
106
Figure 1. Begg’s Funnel Plot With Pseudo 95% Confidence
Limits
2
0
log[OR]
Comparing the relapse rates as a function of time at
follow-up revealed that antidepressants at 3 months of
follow-up significantly reduced relapse rates compared
to placebo (OR = 0.25, 95% CI = 0.17 to 0.36, p < .001)
(Figure 3). At further follow-up, this remained unchanged
at 6 months (OR = 0.19, 95% CI = 0.13 to 0.29, p < .001),
at 9 months (OR = 0.29, 95% CI = 0.21 to 0.40, p < .001),
and at 12 months (OR = 0.27, 95% CI = 0.12 to 0.60,
p = .001) (Figures 4–6). Meta-regression confirmed this
result: with longer follow-up, there was no significant
additive relapse-reducing effect, neither at 6 months
compared to the first 3 months (meta-regression coefficient = –0.24, 95% CI = –0.77 to 0.28, p = .367), nor at 9
months compared to the first 3 months (meta-regression
coefficient = 0.14, 95% CI = –0.44 to 0.72, p = .635),
–2
–4
–6
0
0.5
1
SE of log[OR]
Abbreviation: OR = odds ratio.
1.5
Kaymaz et al.
Figure 2. Recurrence in the Antidepressant and Placebo Groups in the Pooled Analysis of All Included Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Study
Montgomery et al, 198824
Doogan and Caillard, 199225
Montgomery and Rasmussen, 199226
Montgomery and Dunbar, 199327
Robert and Montgomery, 199528
Keller et al, 199829
Reimherr et al, 199830
Terra and Montgomery, 199831
Hochstrasser et al, 200132
Klysner et al, 200233
Mindham et al, 197234
Klerman et al, 197435
Coppen et al, 197836
Stein et al, 198037
Bialos et al, 198238
Kane et al, 1982 39
Glen et al, 198440
Prien et al, 198441
Cook et al, 1986 42
Georgotas et al, 198943
Rouillon et al, 198944
Frank et al, 199045
OADIG, 199346
Reynolds et al, 1999 47
Alexopoulos et al, 200048
Wilson et al, 2003 49
Gilaberte et al, 200150
Schmidt et al, 200051
McGrath et al, 200652
Perahia et al, 200653
Overalla
Odds Ratio (95% CI)
0.26 (0.14 to 0.49)
0.19 (0.11 to 0.35)
0.29 (0.12 to 0.73)
0.25 (0.11 to 0.57)
0.50 (0.25 to 1.01)
0.35 (0.18 to 0.68)
0.12 (0.06 to 0.23)
0.30 (0.15 to 0.60)
0.27 (0.16 to 0.48)
0.23 (0.11 to 0.48)
0.28 (0.11 to 0.70)
0.35 (0.08 to 1.55)
0.51 (0.10 to 2.62)
0.20 (0.06 to 0.63)
0.02 (0.00 to 0.48)
0.28 (0.01 to 8.42)
0.13 (0.01 to 1.52)
0.29 (0.11 to 0.77)
0.14 (0.01 to 3.35)
2.00 (0.35 to 11.36)
0.43 (0.33 to 0.57)
0.08 (0.02 to 0.29)
0.35 (0.13 to 0.94)
0.09 (0.02 to 0.35)
0.20 (0.05 to 0.80)
0.61 (0.29 to 1.31)
0.38 (0.18 to 0.80)
0.35 (0.22 to 0.57)
0.33 (0.20 to 0.56)
0.33 (0.19 to 0.59)
0.30 (0.25 to 0.35)
0.000806
1
% Weight
5.0
5.6
2.9
3.5
4.3
4.6
4.8
4.2
5.7
3.8
2.9
1.2
1.0
2.0
0.3
0.3
0.5
2.6
0.3
0.9
10.3
1.5
2.5
1.4
1.4
3.8
3.8
6.8
6.3
5.7
1240.17
Odds Ratio
a
Pooled odds ratio and CI in random-effects model.
Abbreviations: CI = confidence interval, OADIG = Old Age Depression Interest Group.
nor at 12 months compared to the first 3 months (metaregression coefficient = 0.10, 95% CI = –0.65 to 0.85,
p = .790).
Duration of Continuation Phase
In meta-regression, the prerandomization periods of
1 to 3 months and of > 3 to 6 months after achievement
of remission showed no significantly different relapsereducing effect of the antidepressant compared to a prerandomization treatment period of less than 1 month
(meta-regression coefficient = 0.41, 95% CI = –0.73 to
1.57, p = .478 and meta-regression coefficient = 0.20,
95% CI = –0.26 to 0.68, p = .389, respectively).
Number of Previous Depressive Episodes
The meta-regression comparing patients with recurrent
episode(s) versus those with a single episode revealed
a significant positive regression coefficient, indicating
less antidepressant benefit for patients with recurrent episodes (meta-regression coefficient = 1.6, 95% CI = 0.60
to 2.59, p = .002). Meta-analyses stratified for this variable revealed that the pooled OR for relapse in single episode patients was considerably lower (OR = 0.12, 95%
CI = 0.06 to 0.26, p < .001), without significant heterogeneity (χ2 = 10.79, df = 7, p = .148), compared to the OR
for relapse in recurrent episode patients (OR = 0.37, 95%
CI = 0.31 to 0.44, p < .001), also without significant heterogeneity (χ2 = 23.15, df = 22, p = .393).
Mode of Discontinuation of the Antidepressant
With regard to the question of possible moderation of
the effect size of antidepressant as a function of mode of
discontinuation of the antidepressant, it was found that
the relapse rates in studies with gradual discontinuation of
the antidepressant (i.e., ≥ 1 week) were not different
from those in studies with abrupt discontinuation (i.e., < 1
week) (meta-regression coefficient = –0.56, 95% CI =
–1.37 to 0.24, p = .174).
Interaction Between the Number of Previous Episodes
and Mode of Discontinuation of the Antidepressant
In the meta-regression, there was a significant positive
interaction between the variables number of previous episodes and mode of discontinuation (meta-regression coefficient = 2.15, 95% CI = 1.01 to 3.29, p < .001). Thus, in
the recurrent episode patients, abrupt discontinuation was
107
Antidepressant Discontinuation and Recurrent Episodes
Figure 3. Recurrence in the Antidepressant and the Placebo Groups in the Pooled Analysis at 3 Months of Follow-Up in the
Maintenance Treatment
1
2
3
4
5
6
7
8
9
10
11
17
18
20
21
22
23
24
25
26
27
29
30
Study
Montgomery et al, 198824
Doogan and Caillard, 199225
Montgomery and Rasmussen, 199226
Montgomery and Dunbar, 199327
Robert and Montgomery, 199528
Keller et al, 199829
Reimherr et al, 199830
Terra and Montgomery, 199831
Hochstrasser et al, 200132
Klysner et al, 200233
Mindham et al, 197234
Glen et al, 198440
Prien et al, 198441
Georgotas et al, 198943
Rouillon et al, 198944
Frank et al, 199045
OADIG, 199346
Reynolds et al, 199947
Alexopoulos et al, 200048
Wilson et al, 200349
Gilaberte et al, 200150
McGrath et al, 200652
Perahia et al, 200653
Overalla
Odds Ratio (95% CI)
0.30 (0.14 to 0.67)
0.06 (0.02 to 0.16)
0.18 (0.06 to 0.51)
0.11 (0.02 to 0.54)
0.61 (0.25 to 1.46)
0.17 (0.04 to 0.82)
0.10 (0.05 to 0.23)
0.10 (0.03 to 0.29)
0.18 (0.09 to 0.37)
0.05 (0.01 to 0.17)
0.25 (0.08 to 0.73)
0.29 (0.02 to 3.52)
0.11 (0.03 to 0.39)
8.00 (0.80 to 79.65)
0.49 (0.32 to 0.74)
0.03 (0.00 to 0.60)
0.40 (0.10 to 1.52)
0.32 (0.08 to 1.30)
0.10 (0.00 to 2.04)
0.75 (0.28 to 2.04)
0.34 (0.14 to 0.82)
0.53 (0.32 to 0.90)
0.47 (0.26 to 0.84)
0.25 (0.17 to 0.36)
0.001614
1
% Weight
5.7
5.2
4.7
3.2
5.4
3.2
5.7
4.5
6.0
3.8
4.6
1.7
3.9
1.9
7.1
1.3
3.8
3.7
1.2
4.9
5.3
6.7
6.5
619.752
Odds Ratio
a
Pooled odds ratio and CI in random-effects model.
Abbreviations: CI = confidence interval, OADIG = Old Age Depression Interest Group.
Figure 4. Recurrence in the Antidepressant and the Placebo Groups in the Pooled Analysis at 6 Months of Follow-Up in the
Maintenance Treatment
1
2
3
4
5
6
7
8
9
10
11
17
18
21
22
23
25
26
27
29
30
Study
Montgomery et al, 198824
Doogan and Caillard, 199225
Montgomery and Rasmussen, 199226
Montgomery and Dunbar, 199327
Robert and Montgomery, 199528
Keller et al, 199829
Reimherr et al, 199830
Terra and Montgomery, 199831
Hochstrasser et al, 200132
Klysner et al, 200233
Mindham et al, 197234
Glen et al, 198440
Prien et al, 198441
Rouillon et al, 198944
Frank et al, 199045
OADIG, 199346
Alexopoulos et al, 200048
Wilson et al, 200349
Gilaberte et al, 200150
McGrath et al, 200652
Perahia et al, 200653
Overalla
Odds Ratio (95% CI)
0.27 (0.08 to 0.93)
0.02 (0.00 to 0.13)
0.14 (0.03 to 0.70)
0.16 (0.04 to 0.63)
0.30 (0.12 to 0.77)
0.07 (0.01 to 0.53)
0.25 (0.11 to 0.57)
0.55 (0.07 to 3.98)
0.12 (0.04 to 0.36)
0.02 (0.00 to 0.13)
0.16 (0.03 to 0.74)
0.20 (0.01 to 3.66)
0.33 (0.07 to 1.56)
0.25 (0.15 to 0.41)
0.01 (0.00 to 0.16)
0.11 (0.01 to 1.04)
0.25 (0.02 to 3.10)
1.20 (0.49 to 2.93)
0.13 (0.05 to 0.33)
0.45 (0.20 to 0.99)
0.08 (0.01 to 0.63)
0.19 (0.13 to 0.29)
0.000288
1
Odds Ratio
a
Pooled odds ratio and CI in random-effects model.
Abbreviations: CI = confidence interval, OADIG = Old Age Depression Interest Group.
108
3470.93
% Weight
5.5
3.0
4.2
5.1
7.1
2.9
7.8
3.1
6.4
3.3
4.3
1.7
4.4
9.6
1.5
2.7
2.2
7.3
7.2
7.9
2.9
Kaymaz et al.
Figure 5. Recurrence in the Antidepressant and the Placebo Groups in the Pooled Analysis at 9 Months of Follow-Up in the
Maintenance Treatment
Study
1
2
4
6
7
8
9
10
18
21
23
25
26
27
29
20
Odds Ratio (95% CI)
Montgomery et al, 198824
Doogan and Caillard, 199225
Montgomery and Dunbar, 199327
Keller et al, 199829
Reimherr et al, 199830
Terra and Montgomery, 199831
Hochstrasser et al, 200132
Klysner et al, 200233
Prien et al, 198441
Rouillon et al, 198944
OADIG, 199346
Alexopoulos et al, 200048
Wilson et al, 2003 49
Gilaberte et al, 200150
McGrath et al, 200652
Georgotas et al, 198943
Overalla
0.28 (0.08 to
0.25 (0.09 to
0.94 (0.16 to
0.62 (0.04 to
0.52 (0.20 to
0.36 (0.12 to
0.23 (0.05 to
0.04 (0.00 to
1.00 (0.08 to
0.36 (0.22 to
0.82 (0.14 to
0.04 (0.00 to
0.20 (0.06 to
0.13 (0.05 to
0.18 (0.06 to
(excluded)
0.29 (0.21 to
0.002639
1
0.95)
0.71)
5.50)
10.14)
1.34)
1.11)
1.09)
0.72)
12.56)
0.61)
4.74)
0.48)
0.64)
0.32)
0.57)
% Weight
6.3
8.4
3.1
1.3
10.0
7.3
3.9
1.2
1.5
27.5
3.1
1.6
6.9
10.9
6.9
0.0
0.40)
378.898
Odds Ratio
a
Pooled odds ratio and CI in random-effects model.
Abbreviations: CI = confidence interval, OADIG = Old Age Depression Interest Group.
Figure 6. Recurrence in the Antidepressant and the Placebo Groups in the Pooled Analysis at 12 Months of Follow-Up in the
Maintenance Treatment
Odds Ratio (95% CI)
% Weight
4 Montgomery and Dunbar, 199327
Study
0.07 (0.02 to 0.35)
12.6
9 Hochstrasser et al, 200132
0.18 (0.03 to 1.01)
11.6
21 Rouillon et al, 198944
0.61 (0.31 to 1.17)
21.1
23 OADIG, 199346
0.13 (0.01 to 2.96)
5.2
26 Wilson et al, 2003 49
0.20 (0.06 to 0.64)
16.3
27 Gilaberte et al, 200150
0.13 (0.06 to 0.32)
19.1
29 McGrath et al, 200652
1.58 (0.39 to 6.37)
14.1
6 Keller et al, 199829
(excluded)
Overalla
0.0
0.27 (0.12 to 0.60)
0.005856
1
170.764
Odds Ratio
a
Pooled odds ratio and CI in random-effects model.
Abbreviations: CI = confidence interval, OADIG = Old Age Depression Interest Group.
associated with a lower relapsing preventive effect of
antidepressants (OR = 0.32, 95% CI = 0.27 to 0.38, p <
.001; heterogeneity: χ2 = 57.20, df = 41, p = .048), compared to multiple episode patients with gradual discontinuation (OR = 0.11, 95% CI = 0.06 to 0.21, p < .001;
heterogeneity: χ2 = 8.30, df = 9, p = .504), with nonoverlapping confidence intervals of the ORs in the 2 groups.
In single episode patients, no such large difference was
apparent, and confidence intervals were largely overlapping (abrupt: OR = 0.10, 95% CI = 0.03 to 0.28, p < .001;
gradual: OR = 0.19, 95% CI = 0.04 to 0.87, p = .032). The
significant heterogeneity in the meta-analysis of recurrent
episode patients with abrupt discontinuation was largely
due to 1 small study with an unusually large effect size in
109
Antidepressant Discontinuation and Recurrent Episodes
the opposite direction (Robert and Montgomery28). This
study has a meta-analytic weight of less than 1%; therefore, excluding it has little impact on the effect size (but
does reduce heterogeneity). Exclusion of this study from
the analysis reduced heterogeneity (χ2 = 46.00, df = 39,
p = .205; OR = 0.32, 95% CI = 0.27 to 0.37, p < .001).
DISCUSSION
On the basis of the pooled results of 30 randomized
clinical trials, it can be concluded that treatment with an
antidepressant results in an approximately 70% reduction
of risk of relapse. Thus, the current study confirms the
effectiveness of continuation treatment with antidepressants once remission has been achieved, as was reported
in previous meta-analyses by Loonen et al.,3 Geddes et
al.,6 and Viguera et al.54 In agreement with the latter 2
meta-analyses, it was found that the difference between
active medication and placebo was not greater in the studies with SSRIs compared to the studies with TCAs. However, it should be emphasized that no studies were found
in which SSRIs and TCAs were compared head-to-head
with each other.
In this meta-analysis, several additional questions concerning the continuation and maintenance treatment of
depressive disorder were addressed.
Duration of Continuation Phase
Regarding the first question (of how long treatment
with antidepressants should be continued once remission
has been attained), no significant difference was found
between studies with patients who were randomized
within 1 month after having achieved remission compared
to those in which patients were randomized after a continuation treatment of 1 to 3 months or after 3 to 6 months.
This is in line with the results of previous metaanalyses6,54 that also did not find a difference in relapse
rates in relation to the duration of the stabilization prior to
discontinuation of the antidepressant.
In answering the question of how long antidepressants
should be continued, another perspective can be gained
from the comparison of relapse/recurrence rates over the
follow-up at 3, 6, 9, and 12 months after randomization.
We found significant relapse-reducing effects of antidepressants compared to placebo at 3, 6, and 9 months as
well as 12 months of follow-up. However, the difference
between antidepressant and placebo was already achieved
within 3 months after randomization, with no additional
reduction in risk at 6, 9, and 12 months compared to the
effect already obtained during the previous periods (up to
3, 6, and 9 months, respectively).
Unfortunately, it was not possible to include studies
in which patients were randomized during maintenance
treatment, i.e., more than 6 months after remission was
achieved. The only available data are from 2 small stud-
110
ies. The first was by Bialos et al.,38 who studied 17 patients who had been receiving long-term amitriptyline
treatment. Eight of 10 patients who had their medication
tapered and discontinued had a relapse within 4 months
compared to none of the 7 control subjects during the 6
months of the study. The second small study was by Cook
et al.,42 who studied 15 patients who had been receiving a
long-term treatment with a tricyclic antidepressant agent.
None of the 9 patients who continued on active medication experienced a relapse, whereas 3 of the 9 patients
switched to placebo experienced a relapse. With the exception of these small studies, no other studies have specifically addressed the question of whether treatment with
antidepressants should be continued longer than 6 months
after remission is achieved.
On the basis of the above data taken together, it is not
possible to give recommendations for an optimal duration
of continuation and maintenance treatment with antidepressants. In fact, there is also no evidence from the reviewed studies of the effect of discontinuation of antidepressants to justify the defined distinction between
continuation treatment (up to 6 months) and maintenance
treatment (beyond 6 months).
Number of Previous Depressive Episodes
A history of severe and frequently recurring depressive
episodes is considered to be a plausible clinical predictor
of increased risk of relapse or recurrences after discontinuation of the antidepressant.22,26,55
Regarding the second question (of whether treatment
should be continued longer in patients with recurrent episodes than in patients who have suffered from a first or
second episode), it was found that the reduction in relapse
rates was greater for recurrent episode patients compared
to single episode patients. In fact, the results support previous findings that patients with 1 or more previous depressive episodes have significantly less benefit from the
relapse-reducing effect of the antidepressant than patients
with a first episode.29,54,56 Thus, the results suggest that
with longer duration of illness, the risk of relapse is more
difficult to control, conforming to the sensitization hypothesis proposed by Post et al.7 In addition, the data
showed that the reduction in the protective effect of antidepressants was specifically evident in the subgroup of
patients in which the antidepressant was discontinued
abruptly. Post et al. hypothesized that stress of a particular
type, intensity, and intermittency may produce sensitization in a fashion similar to the behavioral sensitization.
Even in cases of anticipated stresses or imagined losses,
if sufficiently conditioned, the behavioral, physiologic,
and biochemical alterations usually associated with an
affective episode might be produced. It may also explain
how stress-induced mood alterations might become so
sensitized that they also occur spontaneously. However, it
should also be emphasized that it was not possible to
Kaymaz et al.
specifically address the question of whether maintenance
treatment is especially indicated for patients with recurrent
depression, again because there are no studies in which
patients were randomized more than 6 months after having
achieved remission.
Other Factors
One of the possible mechanisms for the high relapse
rates in the first 3 months of randomization may be associated with differences in the half-lives of the various antidepressant agents,62,63 assuming that the shorter the halflife of the antidepressant, the greater the risk of relapse.
However, addressing this question with adequate precision
was not possible, since the range of half-lives of the antidepressants used in the different studies is extremely heterogeneous and in addition imprecise, given a great level
of interindividual variation, depending on whether the
patient is a fast or a slow metabolizer and depending on
gender- and age-specific aspects of the elimination of the
antidepressant.
and maintenance treatment thereafter. We looked at the different guidelines and compared the recommendations with
the findings of this study.
The recommendations in most guidelines64–67 stating
that patients should be treated for at least 6 months after
having achieved remission, and especially that patients
with frequent/more previous episodes should receive
maintenance treatment, cannot be supported on the basis of
the current findings. More specifically, any recommendation about maintenance treatment is not evidence based,
due to the fact that studies in which patients were randomized more than 6 months after having achieved remission
are lacking. The ORs reported in this article indicate that
gradual discontinuation leads to more relapse than abrupt
discontinuation. However, it is not possible to differentiate
to what extent the smaller OR for single episode studies
(OR = 0.12) than for recurrent episode studies (OR = 0.36)
can be explained by more relapse with placebo in single
episode studies or by less relapse with antidepressant medication in single episode studies. Further investigation of
the above mentioned possible pathways for smaller ORs is
warranted before firm conclusions can be drawn. Furthermore, in the group of patients with recurrent depressive
episodes, those with abrupt discontinuation of the antidepressant appeared to benefit less from the relapse-reducing
effect of the antidepressant (OR = 0.32) than those with
gradual discontinuation (OR = 0.11). The interpretation of
this finding is that since the relapse rate during continuing
treatment with an antidepressant should not be different in
these studies, the ORs can arguably only be taken to indicate that gradual discontinuation leads to more relapse in
placebo groups than abrupt discontinuation. This is a
somewhat counterintuitive finding and suggests that findings resulting from models including interaction terms
may be due to chance. Furthermore, gradual withdrawal
can be recommended in order to prevent other withdrawal
symptoms.68–70
Another issue is how to perform continuation or maintenance treatment. Some studies detailing the management
of a relapse or a recurrence point out that patients who did
relapse in the continuation phase after initially having responded to an antidepressant can benefit from an increase
in the dose of the same antidepressant or from an increase
using an enteric-coated antidepressant initially dosed once
a week to twice a week.71,53,63 In the case of a recurrence
after the discontinuation of medication in the maintenance
phase, patients can benefit from a reinstatement of the
antidepressant. We did not address these aspects in our
meta-analysis.
Consequences
Several guidelines on the treatment of patients with depressive disorder have addressed the issue of long-term
treatment with antidepressants: continuation treatment
during the first 6 months after achievement of remission
Limitations
The results of our study should be seen in the light of
several methodological limitations.
Several reports did not specify certain relevant details,
such as the specific antidepressant studied (some reports
Mode of Discontinuation of the Antidepressant
Regarding the third question (of how quickly the antidepressant should be discontinued), it was found that relapse rates in studies in which patients did discontinue
medication abruptly (i.e., < 1 week) were not different
from rates in studies in which patients gradually discontinued their antidepressant (i.e., ≥ 1 week), but that instead
mode of discontinuation was relevant only for the particular subgroup of recurrent episode patients. This subgroup
effect quite likely explains why Viguera et al.54 found that
relapse rates off medication did not differ significantly between studies involving rapid discontinuation and those in
which tapering was more gradual for the total group of different types of antidepressants that they included, as the
interaction with number of previous episodes was not
investigated in that study. Differential distribution of recurrent episode patients may explain why Geddes and
colleagues6 reported an excess of relapse immediately following discontinuation of the antidepressant in the first
month of drug discontinuation. Thus, when the interaction
with recurrent episode patients is modeled, there is clear
evidence that acute withdrawal of medication might induce a relapse, a problem that has also been identified for
lithium, for which acute withdrawal can lead to manic relapses,57–60 and antipsychotics, for which a higher risk of
psychotic relapse was found within 6 months of discontinuation, particularly in hospitalized patients and patients
in whom the antipsychotics were withdrawn abruptly.61
111
Antidepressant Discontinuation and Recurrent Episodes
just mentioned TCA), doses of the antidepressant, the
illness history and more specifically the precise number
of previous depressive episodes, the precise duration of
treatment after achievement of remission and prior to randomization, and in some studies the mode of discontinuation of the antidepressant. Diagnostic criteria varied, and
some studies included an unspecified number of patients
with other disorders such as a depression in the course
of bipolar II disorder, dysthymia, atypical depression/
depression not otherwise specified, or a major depressive
episode with a significant comorbid disorder. When relevant details could not be retrieved, the studies were
excluded from the subanalyses.
Definitions of remission as well as of relapse/
recurrence usually varied between the studies from formal definitions including clinical assessment with application of diagnostic criteria or the use of rating scales,
to more global criteria such as worsening of depressive
symptoms severe enough to warrant hospitalization or reinstitution of antidepressant treatment. Therefore, it remained unclear in some studies to what degree patients
actually had attained full remission prior to randomization and/or to what degree they indeed suffered from a relapse or recurrence of a formal major depressive episode.
The trials were also heterogeneous in terms of diagnostic criteria, dropout rates, the power at the start of
the trial, drugs used, and outcome criteria (Table 1). Although the main meta-analytic result did not display significant heterogeneity, the p value was close to .05, suggesting underlying sources of variance. Some of these
factors were most likely identified in the meta-regression,
in particular the number of previous depressive episodes
and the interaction of this variable with mode of discontinuation. Further exploration of other potential variables
contributing to heterogeneity, such as age, gender, year of
publication, or drug regimen, was not possible due to the
limitations inherent to all meta-analyses performed without access to individual patient data, and potential differences between trials in the definition and validation of
end points as well as the clinical characteristics of the
randomized patients.
Most of the patients participating in the trials consisted
of patients in secondary care settings with a more severe
and often recurrent type of depressive disorder and thus
a high risk of relapse. Patients with milder depressive
disorders, such as patients treated in primary care, were
underrepresented, so that we can make no inference about
the generalizability of our results to this group of patients.
Whether the relapse rates can be explained by the degree of resistance in patients participating in the different
studies remains unclear, since the degree of resistance to
therapy, prior to participation, was not mentioned in the
studies included.
Another limitation is that our results may be subject to
publication bias as negative trials are more likely to re-
112
main unpublished.72 This is a general limitation of any
conclusion based on perusal of the literature. Using a
Begg’s funnel plot, however, it was found that there was
symmetry in the relationship between effect size and
sample size (see Figure 1), with only a slight overrepresentation of small studies with a positive result.
Drug names: citalopram (Celexa and others), duloxetine (Cymbalta),
fluoxetine (Prozac and others), fluvoxamine (Luvox and others),
imipramine (Tofranil and others), nortriptyline (Pamelor, Aventyl,
and others), paroxetine (Paxil, Pexeva, and others), sertraline (Zoloft
and others).
REFERENCES
1. Angst J. Epidemiology of depression. Psychopharmacology (Berl)
1992;106(suppl):S71–S74
2. Spijker J, de Graaf R, Bijl RV, et al. Duration of major depressive episodes in the general population: results from The Netherlands Mental
Health Survey and Incidence Study (NEMESIS). Br J Psychiatry 2002;
181:208–213
3. Loonen AJ, Peer PG, Zwanikken GJ. Continuation and maintenance
therapy with antidepressive agents: meta-analysis of research. Pharm
Weekbl Sci 1991;13:167–175
4. Joffe R, Sokolov S, Streiner D. Antidepressant treatment of depression:
a meta-analysis. Can J Psychiatry 1996;41:613–616
5. Storosum JG, van Zwieten BJ, Vermeulen HD, et al. Relapse and
recurrence prevention in major depression: a critical review of placebocontrolled efficacy studies with special emphasis on methodological
issues. Eur Psychiatry 2001;16:327–335
6. Geddes JR, Carney SM, Davies C, et al. Relapse prevention with antidepressant drug treatment in depressive disorders: a systematic review.
Lancet 2003;361:653–661
7. Post RM, Rubinow DR, Ballenger JC. Conditioning and sensitisation
in the longitudinal course of affective illness. Br J Psychiatry 1986;149:
191–201
8. Frank E, Prien RF, Jarrett RB, et al. Conceptualization and rationale for
consensus definitions of terms in major depressive disorder: remission,
recovery, relapse, and recurrence. Arch Gen Psychiatry 1991;48:851–855
9. STATA Statistical Software: Release 9.0. College Station, Tex:
StataCorp; 2005
10. Kocsis JH, Friedman RA, Markowitz JC, et al. Maintenance therapy for
chronic depression: a controlled clinical trial of desipramine. Arch Gen
Psychiatry 1996 Sep;53(9):769–774; discussion 775–766
11. Franchini L, Gasperini M, Perez J, et al. Dose-response efficacy
of paroxetine in preventing depressive recurrences: a randomized,
double-blind study. J Clin Psychiatry 1998 May;59(5):229–232
12. Franchini L, Zanardi R, Gasperini M, et al. Two-year maintenance treatment with citalopram, 20 mg, in unipolar subjects with high recurrence
rate. J Clin Psychiatry 1999 Dec;60(12):861–865
13. Franchini L, Gasperini M, Zanardi R, et al. Four-year follow-up study
of sertraline and fluvoxamine in long-term treatment of unipolar subjects
with high recurrence rate. J Affect Disord 2000;58:233–236
14. Koran LM, Gelenberg AJ, Kornstein SG, et al. Sertraline versus imipramine to prevent relapse in chronic depression. J Affect Disord 2001;65:
27–36
15. Kishimoto A, Mizukawa R, Matsuzaki R, et al. Prophylactic effect of
mianserin on recurrent depression. Acta Psychiatr Scand 1994;89:46–51
16. Flint AJ, Rifat SL. Recurrence of first-episode geriatric depression after
discontinuation of maintenance antidepressants. Am J Psychiatry 1999;
156:943–945
17. Flint AJ, Rifat SL. Maintenance treatment for recurrent depression in late
life: a four-year outcome study. Am J Geriatr Psychiatry 2000;8:112–116
18. Lepine JP, Caillard V, Bisserbe JC, et al. A randomized, placebocontrolled trial of sertraline for prophylactic treatment of highly
recurrent major depressive disorder. Am J Psychiatry 2004;161:836–842
19. Bjork K. The efficacy of zimeldine in preventing depressive episodes in
recurrent major depressive disorders: a double-blind placebo-controlled
study. Acta Psychiatr Scand Suppl 1983;308:182–189
20. Eric L. A prospective double-blind comparative multi-centre study
Kaymaz et al.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
of paroxetine in preventing recurrent major depressive episodes.
Biol Psychiatry 1991;29:11–14
May M, Veltro F, Pirozzi R, et al. Pattern of recurrence of illness after
recovery from an episode of major depression: a prospective study.
Am J Psychiatry 1992;149:795–800
Kupfer DJ, Frank E, Perel JM, et al. Five-year outcome for maintenance
therapies in recurrent depression. Arch Gen Psychiatry 1992;49:769–773
McGrath PJ, Stewart JW, Petkova E, et al. Predictors of relapse during
fluoxetine continuation or maintenance treatment of major depression.
J Clin Psychiatry 2000 Jul;61(7):518–524
Montgomery SA, Dufour H, Brion S, et al. The prophylactic efficacy
of fluoxetine in unipolar depression. Br J Psychiatry Suppl 1988
Sep;(3):69–76
Doogan DP, Caillard V. Sertraline in the prevention of depression.
Br J Psychiatry 1992;160:217–222
Montgomery SA, Rasmussen JG. Citalopram 20 mg, citalopram 40 mg
and placebo in the prevention of relapse of major depression. Int Clin
Psychopharmacol 1992;6:71–73
Montgomery SA, Dunbar G. Paroxetine is better than placebo in relapse
prevention and the prophylaxis of recurrent depression. Int Clin
Psychopharmacol 1993;8:189–195
Robert P, Montgomery SA. Citalopram in doses of 20–60 mg is effective
in depression relapse prevention: a placebo-controlled 6 month study.
Int Clin Psychopharmacol 1995;10:29–35
Keller MB, Kocsis JH, Thase ME, et al. Maintenance phase efficacy
of sertraline for chronic depression: a randomized controlled trial.
JAMA 1998;280:1665–1672
Reimherr FW, Amsterdam JD, Quitkin FM, et al. Optimal length of
continuation therapy in depression: a prospective assessment during
long-term fluoxetine treatment. Am J Psychiatry 1998;155:1247–1253
Terra JL, Montgomery SA. Fluvoxamine prevents recurrence of depression: results of a long-term, double-blind, placebo-controlled study.
Int Clin Psychopharmacol 1998;13:55–62
Hochstrasser B, Isaksen PM, Koponen H, et al. Prophylactic effect of
citalopram in unipolar, recurrent depression: placebo-controlled study
of maintenance therapy. Br J Psychiatry 2001;178:304–310
Klysner R, Bent-Hansen J, Hansen HL, et al. Efficacy of citalopram
in the prevention of recurrent depression in elderly patients: placebocontrolled study of maintenance therapy. Br J Psychiatry 2002;181:29–35
Mindham RH, Howland C, Shepherd M. Continuation therapy with
tricyclic antidepressants in depressive illness. Lancet 1972;2:854–855
Klerman GL, Dimascio A, Weissman M, et al. Treatment of depression
by drugs and psychotherapy. Am J Psychiatry 1974;131:186–191
Coppen A, Ghose K, Montgomery SA, et al. Continuation therapy
with amitriptyline in depression. Br J Psychiatry 1978;133:28–33
Stein MK, Rickels K, Weise CC. Maintenance therapy with amitriptyline: a controlled trial. Am J Psychiatry 1980;137:370–371
Bialos D, Giller E, Jatlow P, et al. Recurrence of depression after discontinuation of long-term amitriptyline treatment. Am J Psychiatry 1982;
139:325–329
Kane JM, Quitkin FM, Rifkin A, et al. Lithium carbonate and imipramine
in the prophylaxis of unipolar and bipolar II illness: a prospective,
placebo-controlled comparison. Arch Gen Psychiatry 1982;39:
1065–1069
Glen AI, Johnson AL, Shepherd M. Continuation therapy with lithium
and amitriptyline in unipolar depressive illness: a randomized, doubleblind, controlled trial. Psychol Med 1984 Feb;14(1):37–50
Prien RF, Kupfer DJ, Mansky PA, et al. Drug therapy in the prevention
of recurrences in unipolar and bipolar affective disorders: report of the
NIMH Collaborative Study Group comparing lithium carbonate, imipramine, and a lithium carbonate–imipramine combination. Arch Gen
Psychiatry 1984;41:1096–1104
Cook BL, Helms PM, Smith RE, et al. Unipolar depression in the
elderly: reoccurrence on discontinuation of tricyclic antidepressants.
J Affect Disord 1986;10:91–94
Georgotas A, McCue RE, Cooper TB. A placebo-controlled comparison
of nortriptyline and phenelzine in maintenance therapy of elderly
depressed patients. Arch Gen Psychiatry 1989;46:783–786
Rouillon F, Phillips R, Serrurier D, et al. Recurrence of unipolar depression and efficacy of maprotiline [in French]. Encephale 1989;15:527–534
Frank E, Kupfer DJ, Perel JM, et al. Three-year outcomes for
maintenance therapies in recurrent depression. Arch Gen Psychiatry
1990;47:1093–1099
46. Old Age Depression Interest Group. How long should the elderly take
antidepressants? a double-blind placebo-controlled study of continuation/prophylaxis therapy with dothiepin. Br J Psychiatry 1993;162:
175–182
47. Reynolds CF, Frank E, Perel JM, et al. Nortriptyline and interpersonal
psychotherapy as maintenance therapies for recurrent major depression:
a randomized controlled trial in patients older than 50 years. JAMA
1999;281:39–45
48. Alexopoulos GS, Meyers BS, Young RC, et al. Executive dysfunction
and long-term outcomes of geriatric depression. Arch Gen Psychiatry
2000;57:285–290
49. Wilson KC, Mottram PG, Ashworth L, et al. Older community residents
with depression: long-term treatment with sertraline: randomised,
double-blind, placebo-controlled study. Br J Psychiatry 2003;182:
492–497
50. Gilaberte I, Montejo AL, de la Gandara J, et al. Fluoxetine in the
prevention of depressive recurrences: a double-blind study. J Clin
Psychopharmacol 2001;21:417–424
51. Schmidt ME, Fava M, Robinson JM, et al. The efficacy and safety
of a new enteric-coated formulation of fluoxetine given once weekly
during the continuation treatment of major depressive disorder. J Clin
Psychiatry 2000 Nov;61(11):851–857
52. McGrath PJ, Stewart JW, Quitkin FM, et al. Predictors of relapse
in a prospective study of fluoxetine treatment of major depression.
Am J Psychiatry 2006;163:1542–1548
53. Perahia DG, Gilaberte I, Wang F, et al. Duloxetine in the prevention of
relapse of major depressive disorder. Br J Psychiatry 2006;188:346–353
54. Viguera AC, Baldessarini RJ, Friedberg J. Discontinuing antidepressant
treatment in major depression. Harv Rev Psychiatry 1998;5:293–306
55. Depression Guideline Panel. Depression in Primary Care, vol 2: Treatment of Major Depression. Rockville, Md: Agency for Health Care
Policy and Research, US Department of Health and Human Services;
1993
56. Ramana R, Paykel ES, Cooper Z, et al. Remission and relapse in major
depression: a two-year prospective follow-up study. Psychol Med 1995;
25:1161–1170
57. Faedda GL, Tondo L, Baldessarini RJ, et al. Outcome after rapid
vs gradual discontinuation of lithium treatment in bipolar disorders.
Arch Gen Psychiatry 1993;50:448–455
58. Suppes T, Baldessarini RJ, Faedda GL, et al. Discontinuation of maintenance treatment in bipolar disorder: risk and implications. Harv Rev
Psychiatry 1993;1:131–144
59. Baldessarini RJ, Tondo L, Faedda GL, et al. Effects of the rate of discontinuing lithium maintenance treatment in bipolar disorders. J Clin
Psychiatry 1996 Oct;57(10):441–448
60. Baldessarini RJ, Tondo L, Floris G, et al. Reduced morbidity after
gradual discontinuation of lithium treatment for bipolar I and II disorders: a replication study. Am J Psychiatry 1997;154:551–553
61. Viguera AC, Baldessarini RJ, Hegarty JD, et al. Clinical risk following
abrupt and gradual withdrawal of maintenance neuroleptic treatment.
Arch Gen Psychiatry 1997;54:49–55
62. Fava M, Mulroy R, Alpert J, et al. Emergence of adverse events following discontinuation of treatment with extended-release venlafaxine.
Am J Psychiatry 1997;154(12):1760–1762
63. Fava M, Detke MJ, Balestrieri M, et al. Management of depression relapse: re-initiation of duloxetine treatment or dose increase. J Psychiatr
Res 2006;40(4):328–326
64. National Institute for Clinical Excellence. Management of Depression in
Primary and Secondary Care. Clinical Guide 23: National Collaborating
Centre for Mental Health; 2004. Available at: http://www.nice.org.uk/
guidance/index.jsp?action = byID&r = true&o = 10958. Accessibility
verified May 16, 2008
65. American Psychiatric Association. Practice Guideline for the Treatment
of Patients With Major Depressive Disorder (Revision). Am J Psychiatry
2000 Apr;157(4 suppl):1–45
66. Bauer M, Whybrow PC, Angst J, et al. World Federation of Societies
of Biological Psychiatry (WFSBF) Task Force on Treatment Guidelines
for Unipolar Depressive Disorders, pt 2: maintenance treatment of major
depressive disorder and treatment of chronic depressive disorders and
subthreshold depressions. World J Biol Psychiatry 2002;3:69–86
67. Kennedy SH, Lam RW, Cohen NL, et al. CANMAT Depression Work
Group. Clinical guidelines for the treatment of depressive disorders,
4: medications and other biological treatments.
113
Antidepressant Discontinuation and Recurrent Episodes
Can J Psychiatry 2001 Jun;46(suppl 1):38S–58S
68. Dilsaver SC, Greden JF. Antidepressant withdrawal phenomena.
Biol Psychiatry 1984;19:237–256
69. Rosenbaum JF, Zajecka J. Clinical management of antidepressant
discontinuation. J Clin Psychiatry 1997;58(suppl 7):37–40
70. Schatzberg AF, Haddad P, Kaplan EM, et al. Serotonin reuptake inhibitor discontinuation syndrome: a hypothetical definition.
114
Discontinuation Consensus Panel. J Clin Psychiatry 1997;58(suppl 7):
5–10
71. Schmidt ME, Fava M, Zhang S, et al. Treatment approaches to major
depressive disorder relapse, pt 1: dose increase. Psychother Psychosom
2002;71(4):190–194
72. Sterne JA, Egger M. Funnel plots for detecting bias in meta-analysis:
guidelines on choice of axis. J Clin Epidemiol 2001;54:1046–1055
CHAPTER 4
Prodromal studies of psychosis
115
DSM-5 and the ‘Psychosis Risk Syndrome’
DSM-5 and the ‘Psychosis Risk Syndrome’: Babylonic Confusion
Nil Kaymaza,b and Jim van Osa,c
aMaastricht University Medical Centre, South Limburg Mental Health Research and Teaching
Network, EURON, Maastricht, The Netherlands. Email: [email protected]; bMediant
GGZ/Mental Health Care, Postbus 775, 7500 AT Enschede, The Netherlands; cDivision of
Psychological Medicine, Institute of Psychiatry, King’s College London, De Crespigny Park,
Denmark Hill, London, UK
[email protected]
Taylor
2010
Jimvan
0000002010
&
Os
Francis
(Received
RPSY_A_475845.sgm
Psychosis:
10.1080/17522431003753233
1752-2439
Original
and
Article
28
Psychological,
(print)/1752-2447
Francis
February
Ltd 2010;
Social
final
(online)
and
version
Integrative
received
Approaches
4 March 2010)
The notion that the important paradigm shift in clinical psychiatry, associated with
early intervention, should now become a diagnostic issue is misguided. The ‘risk’
in Psychosis Risk Syndrome will not make sense to treatment-seeking patients
with distressing symptoms and real need for care, is based on the notion that
labeling people with invalid diagnostic terms has more clinical relevance than
simply addressing care needs, is contingent on elusive sampling strategies posing
as precise diagnostic criteria, and is associated with a false-positive rate of at least
90% in the year after diagnosis. In the 21st century, opinion-based diagnostics
continues to pose a threat to the process of diagnostic revision.
Keywords: Psychosis; Diagnosis; Risk; Schizophrenia; DSM-5
Introduction
DSM-5, for the first time in the history of psychiatry, is considering the inclusion of
a Psychosis Risk Syndrome as a category (www.dsm5.org). The good thing is that the
proposal stems from increased efforts to intervene early in psychotic disorder. Early
intervention improves subsequent course and outcome given that psychotic disorders,
arising during adolescence and early adulthood, disrupt important maturational tasks
and increase the risk of long term exclusion from meaningful societal participation.
The international efforts that have been developed towards intervening earlier during
the initial manifestations of psychotic illness have had a major impact, and are an
example of how outstanding early pioneering work can grow into a true paradigm shift
in clinical psychiatry (Cannon et al., 2008; McGlashan et al., 2006; McGorry et al.,
2002). This exceptional work serves as an example to us all.
There is also a downside however. The same enthusiasm that created the paradigm
of early intervention is now used to turn early intervention into a diagnostic issue,
based on flawed reasoning. The word “risk” is now being used in the Psychosis Risk
Syndrome proposal conceptually has nothing to do with risk, and more to do with
labeling and elusive sampling strategies. Furthermore, no additional diagnostic
category is required: DSM already has a category where early psychotic states can be
classified. Below, these issues will be discussed in more detail.
If one already is help-seeking and in need of care, how can one still be
at risk of something?
The proposed criteria for the Psychosis Risk Syndrome stipulate that “symptoms are
sufficiently distressing and/or disabling to the patient and/or others to lead to helpseeking”. This raises several interesting issues. The first issue is to do with the concept
of risk. If individuals already have reached services and are in need of care, how can
it be that they are still at risk of something? Clearly these individuals are “cases in
116
Psychosis
need of care”, therefore they are no longer “at risk” because they already have reached
a clinical outcome. In DSM terms, they fulfil criteria for Psychotic Disorder, Not
Otherwise Specified (NOS). In fact, the NOS category is ideal as it is commonly used
to describe psychotic states early on in the diagnostic process, when it is not clear how
best to classify symptoms. As a patient’s condition becomes better understood with
the passage of time, the “NOS” can often be replaced by a more precise category. It
is highly confusing for a patient with specific psychotic symptoms, which do not
appear “attenuated” to him, and for which he is seeking treatment, to hear that his
diagnosis is that he is at risk of something: the patient does not feel at risk, he feels in
need of treatment for real problems.
The argument of the Psychosis Risk Syndrome advocates is that the “thing” these
individuals are at risk for is a diagnostic label such as schizophrenia: they may be in
need of treatment, but they do not meet criteria for schizophrenia. For example, in
the North American Multicentre study, 56% of treatment-seeking high risk individuals “converted” to schizophrenia (Cannon et al., 2008; Woods et al., 2009). It could
be argued, however, that a (ICD, RDC, DSM) diagnosis of schizophrenia represents
an odd outcome for risk defined as a state of symptomatic treatment-seeking.
Psychotic disorders, spread over a myriad of diagnostic categories in the different
classification systems, in reality represent highly variable syndromal clusters of
continuous psychotic and affective symptom dimensions, in combination with variable degrees of motivational and neurocognitive impairments (van Os & Kapur,
2009). The notion that a person may be at risk for a specific diagnostic category such
as schizophrenia only makes sense if (ICD, DSM, RDC) diagnoses like “schizophrenia” represent valid nosological entities, i.e. “exist”. As the scientific evidence for
the hypothesis of valid nosological entities is very weak (van Os & Kapur, 2009), it
is difficult to see how any person may be helped by being called “at risk” of meeting
criteria for labels that do not represent disease entities. The obsession with arbitrary
diagnostic labels in psychiatry apparently has taken on such proportions that a
patient with real symptoms and real need for care now must remain in the category
“at risk” as long as he does not meet criteria for a non-existent and non-relevant
diagnostic label.
How “risky” is the risk, and how replicable is it?
The reason the criterion of distress and treatment seeking was included in the Psychosis
Risk Syndrome is because of a very interesting, and hitherto unresolved issue. Attenuated psychotic symptoms are weak predictors of transition to psychotic disorder in
unselected general population samples, but become strong predictors of psychotic
disorder in selected samples of help-seeking individuals in the ultra-high risk literature.
For example, Hanssen and colleagues studied the outcome of psychotic experiences
with onset in the past year, which conform to the criterion of the Psychosis Risk
Syndrome (symptoms must have begun in or significantly worsened in the past year)
(Hanssen, Bak, Bijl, Vollebergh, & van Os, 2005). A two-year follow-up of 79 of these
individuals revealed that 6 converted to clinical psychotic disorder. The one-year
incidence rate therefore is 3.8% (6/158 person-years). A yearly risk of 3.8% is not
negligible, but cannot be considered very useful for the purpose of prediction, particularly as longer term follow-up studies of up to 25 years have shown that risk of
conversion is spread evenly over the follow-up period (Werbeloff et al., 2009). The
ultra-high risk literature, however, displays a different trend. For example, in the
117
DSM-5 and the ‘Psychosis Risk Syndrome’
Multisite North-American study (Cannon et al., 2008), 82 out of 291 patients converted
over a period of 2.5 years, resulting in an incidence rate of 11.3%, around three times
higher than the Hanssen study. How to explain these differences? Clearly, the
difference must reside in the way the samples were collected and “enriched” with risk.
Hanssen and colleagues conducted a follow-up of a representative general population
sample, whereas ultra-high risk samples invariably are opportunistic, each study using
a different, unique strategy that cannot be replicated. For example, the sampling strategy of the ultra-high risk North-American study was described as: “Each site recruited
potential subjects through clinical referrals as stimulated by talks to school counselors
and mental health professionals in community settings”.
Therefore, two questions arise. First, is a yearly risk of around 10% something that
needs to be diagnosed as a “risk syndrome”? For example, the yearly risk for any
person in the general population to develop any kind of psychiatric disorder is around
5–10%, depending on how inclusive the list of disorders is. Clearly, we are all at risk
at any moment during our lives. Is it necessary to specifically label individuals with a
10% risk, even if it concerns a severe disorder? Have we become so health conscious
that we take it for granted that 90% of those diagnosed with the Psychosis Risk
Syndrome in fact are not at risk in the year after the diagnosis?
The second question is also important: How is it possible to introduce a new
diagnostic category that essentially is the product of a non-replicable elusive
sampling strategy occasioning a 300% enrichment in risk, compared to representative general population samples? Surely the diagnostic criteria should also stipulate
how clinicians should go about collecting a sample with attenuated psychotic
symptoms in such a way that the result is an enriched group yielding 10% conversion per year, instead of the much lower rate that they can expect from individuals
with attenuated symptoms in the general population? In other words, the “risk” in
the Psychosis Risk Syndrome is contingent on a sampling enrichment strategy that
remains indefinable. However, the criteria of the Psychosis Risk Syndrome misleadingly make no mention of this.
Conclusion
In conclusion, the appearance of Psychosis Risk Syndrome in the draft of DSM-5
exemplifies how diagnostic revisions in psychiatry remain at risk of opinion-based
rather than evidence-based changes.
Acknowledgements
Nil Kaymaz was supported by the Netherlands Organisation for Scientific Research (NWO),
project number: 017.002.048. Jim van Os is member of the APA DSM-V Psychotic Disorders
Work Group. The views expressed are his own.
References
Cannon, T.D., Cadenhead, K., Cornblatt, B., Woods, S.W., Addington, J., Walker, E.,
Seidman, L.J., Perkins, D., Tsuang, M., McGlashan, T., & Heinssen, R. (2008). Prediction of psychosis in youth at high clinical risk: a multisite longitudinal study in North
America. Arch Gen Psychiatry, 65(1), 28–37.
Hanssen, M., Bak, M., Bijl, R., Vollebergh, W., & van Os, J. (2005). The incidence and
outcome of subclinical psychotic experiences in the general population. Br J Clin Psychol,
44(Pt 2), 181–191.
118
Psychosis
McGlashan, T.H., Zipursky, R.B., Perkins, D., Addington, J., Miller, T., Woods, S.W.,
Hawkins, K.A., Hoffman, R.E., Preda, A., Epstein, I., Addington, D., Lindborg, S., Trzaskoma, Q., Tohen, M., & Breier, A. (2006). Randomized, double-blind trial of olanzapine
versus placebo in patients prodromally symptomatic for psychosis. Am J Psychiatry,
163(5), 790–799.
McGorry, P.D., Yung, A.R., Phillips, L.J., Yuen, H.P., Francey, S., Cosgrave, E.M.,
Germano, D., Bravin, J., McDonald, T., Blair, A., Adlard, S., & Jackson, H. (2002).
Randomized controlled trial of interventions designed to reduce the risk of progression to
first-episode psychosis in a clinical sample with subthreshold symptoms. Arch Gen
Psychiatry, 59(10), 921–928.
van Os, J., & Kapur, S. (2009). Schizophrenia. Lancet, 374(9690), 635–645.
Werbeloff, N., Drukker, M., Dohrenwend, B.P., Levav, I., Yoffe, R., Van Os, J., Davidson,
M., & Weiser, M. (2009). Self-reported psychotic symptoms in the community are
associated with increased risk of later hospitalization for non-affective psychotic disorders
(Conference Abstract). Schizophr Bull, 35(Supplement 1), 74.
Woods, S.W., Addington, J., Cadenhead, K.S., Cannon, T.D., Cornblatt, B.A., Heinssen, R.,
Perkins, D.O., Seidman, L.J., Tsuang, M.T., Walker, E.F., & McGlashan, T.H. (2009).
Validity of the prodromal risk syndrome for first psychosis: findings from the North
American Prodrome Longitudinal Study. Schizophr Bull, 35(5), 894–908.
119
The case of the missing evidence: Do psychotic experiences predict
clinical outcomes in unselected population-based samples? A
systematic review and meta-analysis, enriched with new results.
1, 2
1
Kaymaz N MD , Lataster T PhD , Lieb R PhD
1
ker M PhD
3,4
, Wittchen H-U PhD
3,5
1, 6
, Werbeloff? , Weiser?, Van Os, J PhD , Druk-
1
Department of Psychiatry and Psychology, South Limburg Mental Health Research and Teaching Network, EURON,
Maastricht University Medical Centre, PO Box 616 (DRT 12), 6200 MD Maastricht, The Netherlands
2
Mediant GGZ / Mental Health Care, Enschede, the Netherlands
3
Max Planck Institute of Psychiatry, Kraepelinstrasse 2, D-80804 Munich, Germany.
4
University of Basel, Department of Psychology, Division of Epidemiology and Health Psychology, Missionsstrasse
60-62, 4055 Basel, Switzerland
5
Institute of Clinical Psychology and Psychotherapy, Technical University Dresden, Chemnitzerstr. 46, 01187 Dresden, Germany.
6
King's College London, King's Health Partners, Department of Psychosis Studies, Institute of Psychiatry, London, UK
Correspondence: Jim van Os, Dept. Psychiatry and Psychology, Maastricht University Medical Centre, PO BOX 616,
(DRT12), 6200MD Maastricht, The Netherlands. Tel 31-43-3875443; Fax 31-43-3875444. Email: [email protected]
121
Abstract
Objectives: The reported 10%-20% yearly conversion rate from psychotic experience to clinical psychotic outcome forms the major rational for the practice of intervention in individuals
at ultra high-risk, and the proposed DSM-5 Psychosis Risk Syndrome. However, the base rate
of transition from subthreshold psychotic experience (the exposure) to clinical psychotic
disorder (the outcome) in unselected representative population-based samples is unknown.
Methods: A systematic review and meta-analysis was conducted of representative, longitudinal population-based cohorts with baseline assessment of subclinical psychotic experiences
and follow-up assessment of psychotic and non-psychotic clinical outcomes.
Results: Six cohorts were identified with 3-24 year follow-up of baseline psychotic experiences. The yearly risk of conversion to a clinical psychotic outcome in exposed individuals
(0.56%) was 3.5 times higher than for individuals without psychotic experiences (0.16%).
Conversion risk increased with the number, certainty, frequency, persistence and degree of
affective dysregulation of psychotic experiences. There was also evidence of specificity, as
psychotic experiences only weakly predicted non-psychotic clinical outcomes.
Conclusions: Subclinical psychotic experiences index psychometric risk for psychotic disorder.
However, the discrepancy between the 10% conversion rate in the high-risk literature and
the 0.56% conversion base rate in unselected population-based samples points to the crucial
influence of sample enrichment strategies through community awareness campaigns and
other selective alterations of the permeability of filters on the pathway to mental health care
that lie at the heart of all ultra high-risk studies. Sample enrichment strategies, rather than
clinical high-risk criteria per se, may be the critical factor determining the success of high-risk
early intervention initiatives.
Key words: Psychotic Disorders, Risk, Meta-analysis, Prevention, Delusions, Hallucinations
122
Introduction
Psychotic disorders typically affect young people, disrupting critical maturational tasks. This
fact, together with the observation that longer duration of untreated psychosis predicts
poorer outcome, provides a powerful rational for intervening as early as possible1-3. Given
the fact that psychotic disorders are usually preceded by sometimes lengthy prodromes,
attempts have been made to intervene when individuals are in the early throws of a psychosis prodrome or ‘at risk mental state’4, 5. However, early intervention based on risk can only
be considered productive if samples are sufficiently enriched with cases that will actually
make the conversion to a clinical psychotic disorder in a reasonably short period of time 6, 7.
For example, if the risk of conversion in an ‘at-risk mental state’ sample is 5% over the course
of a year (i.e. at least 100 times higher than the incidence of psychotic disorder in the general
population), 95% of the sample would be treated unnecessarily if the sample underwent
preventive pharmacological or psychotherapeutic treatment. Even if the treatment would be
50% effective in preventing conversion to a psychotic disorder (i.e. half of the 5% would be
prevented from making a conversion), the number needed to inconvenience to prevent one
case of conversion would be 40 (calculated as: 100/2.5), arguably too high for cost-effective
use of sparse resources.
In order to shift the above mathematical considerations to more favourable values, a
considerable amount of literature now exists on attempts to enrich samples in such a way
that the risk of conversion is high enough for effective intervention8. Sample enrichment
strategies may include advertising, increasing awareness in the community and targeting
teachers and primary care physicians. Sample enrichment strategies are not usually described in detail so that standardized approaches have not been developed. For example, the
sample enrichment strategy of the multisite North-American study on prediction of psychosis
was described as: “Each site recruited potential subjects through clinical referrals as stimulated by talks to school counselors and mental health professionals in community settings”9,
whereas in the European Prediction of Psychosis Study10 it was described as: “Knowledge
about early warning signs (eg, concentration and attention disturbances, unexplained functional decline) and inclusion criteria was disseminated (through local workshops, articles in
professional journals and newsletters, informational flyers, and Web sites) to mental health
professionals as well as institutions and persons who might be contacted by at-risk persons
seeking help”. Using these unspecified enrichment strategies, and combining them with criteria on presence of “attenuated” or “transitory” psychotic experiences, family history and
decline in functioning11 initially appeared to yield samples with high conversion rates of up to
54% over 12 months8. However, more recent and more systematically conducted studies in
North America and Europe concur in finding 12-month cumulative conversion rates of 11.3%
(82 out of 291 patients over a period of 2.5 years);9 and 10.1% (37 out of 245 subjects over
18 months);10 respectively.
A major and hitherto unresolved issue regarding the interpretation of these studies is
that the base risk of conversion given expression of subclinical psychotic experiences in the
general population is not known. Therefore, to date it remains unknown to what degree the
10% conversion rate in enriched samples is due to (i) non-standardized and therefore nonreplicable sample enrichment strategies including awareness and education campaigns, (ii)
specific criteria involving family history or decline in functioning and (iii) presence of subclini-
123
cal psychotic experiences6, 12. In addition, research suggests that the risk for non-psychotic
outcomes may also be increased, but non-psychotic disorders as an outcome of subclinical
psychotic experiences are rarely reported6.
In order to address this issue, we reviewed the literature on the risk of developing psychotic disorder given earlier expression of subclinical psychotic experiences in representative, general population samples. To this end, the method of systematic review and metaanalysis was used, as these generally provide a transparent and quantitative approach to
identify, summarise and critically appraise relevant studies, enabling an integrated presentation of results. Furthermore, systematic review and meta-analysis can address metahypotheses over and above primary studies by quantitative exploration of the patterns of
results from single investigations. Specific aims of this meta-analysis were: (i) to examine the
risk of conversion to psychotic disorder given presence of subclinical psychotic experiences in
representative general population samples, (ii) to examine the risk of conversion to nonpsychotic disorder given presence of subclinical psychotic experiences in representative general population samples, (iii) to examine which symptom factors moderate risk of conversion.
In order to achieve these goals, methodology for systematic review as laid down in Metaanalysis of Observational Studies in Epidemiology (MOOSE) guidelines13 was applied. For
some studies, additional analyses were conducted in the original data in order to be able to
also report non-psychotic outcomes that had not been included in the original publications.
Methods
A computerized search strategy was developed to sensitively query the MEDLINE, PsycINFO
and EMBASE databases to identify potentially relevant articles in English, Spanish, German,
French or Dutch, published from 1951 to April 2010. A sensitive search string was composed
with medical subject headings, specific for each database, for concepts relating to “subclinical psychotic experience”, “psychotic disorder” and “conversion/transition”. Reference lists
of articles thus identified were screened in order to encounter additional articles. In addition,
a process of forward and backward citation tracking was executed using the Web of Science
database. Finally, researchers with expertise in the field were contacted in order to identify
additional data potentially relevant for the meta-analysis. Thus, original individual participant
data from two cohort studies, the NEMESIS study14, 15 and the EDSP study16, 17, were subjected to additional analyses in order to add information on non-psychotic outcomes, not
published before, to the meta-analysis database.
In order to reduce methodological variations of studies to be entered in the metaanalysis, a priori criteria for inclusion were formulated. Thus, data of published studies were
added to the meta-analysis database if they (i) were published in a peer-reviewed journal
after 1950; (ii) were written using the English, Spanish, French, German or Dutch language;
(iii) represented a population-based follow-up study of individuals with and without a defined measure of subclinical psychotic experiences at baseline (the exposure) and (iv) provided cumulative incidence rates (or data allowing computation of these) of defined psychotic disorder outcomes (the outcome).
Two investigators independently screened citations and selected publications for further
consideration on the basis of consensus, using three consecutive filters. The first selection
124
filter was at the level of citations, applying the broad criterion of relevance for the topic of
the meta-analysis. The second selection filter was applied at the level of abstracts, excluding
studies that did not meet a single inclusion criterion as defined above. The final filter was
based on inspection of full-text articles. In the case of multiple reports involving a single
study population, the publication with the largest sample size and/or the longest follow-up
was selected.
In the next step, two investigators independently extracted quantitative and qualitative
data from each selected publication. Qualitative data included factors impacting on internal
validity such as methodological and design features as well as the potential for confounding
as well as bias due to differential attrition or possible differential assessment of exposure
and/or outcome. Quantitative data included cumulative conversion rates as a function of
baseline exposure status (i.e. with and without subclinical psychotic experiences). Data pertaining to studies using continuous exposure measures were extracted and analysed according to the original continuous exposure format, and additionally analysed as a dichotomous
exposure in order to facilitate comparison of results across studies. Dichotomisation was
carried out by contrasting, in the case of three-level exposure variables (e.g. no symptoms,
weak symptom, strong symptom), the highest category versus the lowest two. In the case of
four-level exposures, the highest category was similarly compared to lowest three. In case
data could not be extracted in a format suitable for meta-analysis, authors were contacted
for re-analysis of the original data in the required format.
Table 1A was compiled to provide a descriptive summary of selected studies, showing,
for both psychotic and non-psychotic outcomes, the principal study characteristics including
populations, observation periods, exposure and outcome definitions, main results and sample and study design features. Tables 1B and 1C show the quantitative data extracted from
each study for psychotic and non-psychotic outcomes, respectively.
Approach to meta-analysis
Data from the selected studies were combined to estimate pooled rates with their corresponding 95% Confidence Intervals (CIs) under a random effects model, assuming that true
effects were randomly distributed around the mean effect size. The random effect model
presumes that variation in samples and design factors will occasion different true effect sizes
across studies and represented a valid a priori choice given the fact that methods and populations across studies did not correspond to a degree that they could be regarded as estimating the same underlying effect. The between-study variance in the random effects model
reflects heterogeneity across studies, the magnitude of which was evaluated using a chisquare test for heterogeneity, testing whether individual studies varied more than could be
explained by chance alone. In the phase of reading and comparing the articles, various hypotheses for heterogeneity were identified. In case significant heterogeneity was encountered, possible causes were explored by analysing subgroups as well as by including variables
occasioning heterogeneity (modifiers) in meta-regression analyses.
125
Additional analyses undertaken in original datasets
For the specific purpose of the meta-analysis, exposure and outcome data as reported in the
NEMESIS study by Hanssen and colleagues18 and the EDSP study by Dominguez and colleagues19 were subjected to additional analyses. Both the NEMESIS and EDSP studies followed general population cohorts, interviewing the entire cohort with the CIDI on three occasions (EDSP: T0, T2 and T3; NEMESIS: T0, T1 and T2) over time14, 15, 16, 17. In the NEMESIS
study, fresh analyses were conducted to provide additional risk estimates for prevalent exposure (defined as lifetime report of subclinical psychotic experiences at T0, as described in20)
in addition to the incident exposure reported in the original paper18. In both the NEMESIS
and the EDSP data sets, additional analyses were conducted with the following non-psychotic
outcomes: T2 (NEMESIS) or T3 (EDSP) CIDI diagnosis of bipolar disorder, excluding individuals
with a similar diagnosis at T0/T1 (Nemesis) or T0/T2 (EDSP)21-24; T2 (NEMESIS) or T3 (EDSP)
CIDI diagnosis of depressive disorder, excluding individuals with a similar diagnosis at T0/T1
(NEMESIS) or T0/T2 (EDSP) and individuals with a T2 (NEMESIS) or T3 (EDSP) CIDI diagnosis of
bipolar disorder21-24; T2 (NEMESIS) or T3 (EDSP) CIDI diagnosis of anxiety disorder, excluding
individuals with a similar diagnosis at T0/T1 (NEMESIS) or T0/T2 (EDSP) as well as individuals
with a T2 (NEMESIS) or T3 (EDSP) CIDI diagnosis of depressive disorder or bipolar disorder 14,
15, 25
. All extra analyses were conducted in strict accordance with the methodology described
in the original studies and therefore are not reported in detail again here (details available
upon request).
Statistical analysis
All analyses were performed using Stata 1126. A data file including data pertaining to both
psychotic and non-psychotic outcomes was constructed. One study reported 5 different
psychotic outcomes27 – these were combined into a single psychotic outcome.
First, dichotomised exposures were analysed. For these analyses, each record in the data
included sample size, number of subjects with a negative outcome, years of study follow up
and information on modifiers. Using the first three variables, rates per 100,000 person years
were calculated. For each study, at least two records were filled (exposed, non-exposed).
More records were used when rates were stratified by possible outcome modifiers (psychotic/non-psychotic disorder, hospital admission yes/no, incident/prevalent exposure, see
below). Bar charts were generated to present the rates per study for psychotic as well as
non-psychotic outcomes28, 29.
A meta-analysis stratified by exposure and by type of outcome (psychotic/non-psychotic)
generated forest-plots (Stata METAN command). Subsequently, the rates were analysed
using meta-regression analysis (Stata METAREG command). As rates are not normally distributed, and because the number of studies was small and heterogeneity was expected, metaregression was repeated using 1000 permutations (Stata METAREG command with option
PERMUTE). In addition, meta-regression analyses were repeated for more homogeneous
subsets of studies (as described below).
Second, in order to study dose-response effects, exposures were analysed as three-level
variables including 3 categories of graded severity/frequency where available; if there were 4
categories, the two lowest categories were combined to create a similar three-level exposure
126
variable. For this analysis, rates were also presented in a figure and meta-regression analysis
was performed.
Results
Search results
The search yielded 7 articles with data that were pertinent to the meta-analysis as specified
in the criteria above. One study30 was excluded as it concerned a subgroup of persons included in the study by Chapman and colleagues27, already included in the meta-analysis.
The characteristics of the 6 studies, included in the meta-analysis, are listed in table 1A.
All studies had general population sampling frames (two birth cohorts31, 32, three representative general population cohorts18, 19, 33) although in one study the selection was limited to
undergraduate students27; follow up varied from 3 years to 24 years. All studies reported on
variably defined psychotic experiences in the general population and the rate of transition to
variably defined psychotic and non-psychotic clinical outcomes. One study used high level of
schizotypy as predictor27, all other studies used CIDI or related measures of psychotic experiences. Non-psychotic outcomes were depression, mania, anxiety disorder and admission to
hospital for non-psychotic disorder. Some studies described various exposure subgroups,
varying from classification on the basis of number of symptoms (no symptom, single symptom, multiple symptoms); frequency/certainty of psychotic symptom (no symptom, “weak”
symptom and “strong” symptom); psychopathological context (no symptom, symptom without depression, symptom with depression); degree of persistence over 5 years (present at
none, one, two or three assessments over 5 years). Some studies also described rates as a
function of combinations of subgroups (e.g. multiple symptoms with and without comorbid
depression18). All 6 studies reported on psychotic clinical outcomes, and 5 studies additionally reported on other, non-psychotic, clinical outcomes. For one study18, measures of both
incident (psychotic experiences with first onset in the previous year) and prevalent (lifetime
presence of psychotic experiences) exposure were available. One study33 reported servicebased clinical outcomes, defined as admission to hospital.
127
128
3-year longitudinal general
population cohort study
(NEMESIS) assessing psychotic
symptoms and psychiatric
disorders at three time points
(baseline, year 1 and year3)
Hanssen et al.,
2005
11-26 years
41-43 years
Baseline n=1019
FU n=972 (95%)
Risk set: n= 761
T0a: 7076
T1: 5618 (79 %)
T2: 4848 (68 %)
Risk set for analysis: 4042
1. T1 Incident CIDI selfreported CIDI psychotic
symptom, plus:
a) single vs. multiple
b) with vs. without depression
2) T0 Lifetime CIDI selfreported psychotic symptom,
plus:
a) single vs. multiple
c) with vs. without depression
Self-reported psychotic DISC-C
symptom:
1) No-symptom
2) Weak symptom (score 1=
likely)
3) Strong symptom (score 2=
definitely)
T2 diagnosis of Clinical
Psychotic Disorder based on
BPRS severity of psychosis and
CAN need for care.
DSM-IV Schizophreni-form
disorder
DSM-III -R diagnosis:
1) Schizophrenia
2) Psychosis Not Otherwise
Specified
3) Delusional disorder
4) Psychotic bipolar disorder
5) Psychotic depression
Scoring ≥ 1.96 SD on any of the
four scales (exposed, n=375)
and < 0.5 SD above the mean
on each scale (159, nonexposed).
Birth cohort assessed at age 11
years for presence of delusions
and hallucinations and at age
26 years for presence of
psychiatric disorder
20-30 years
(see below)
DSM-III diagnosis:
1) Mania
2) Depression
3) Anxiety disorder
DSM-III-R: diagnosis:
1) Mania/ Bipolar disorder
2) Depression
3) Hypomania
FOLLOW-UP
CIDI interviews by trained
interviewers and telephone
clinical re-interview by clinician
for persons scoring positive on
CIDI psychosis items; BPRS and
need for care scored by
clinician
BASELINE
CIDI interview at all three
measurement points by
trained interviewers plus
clinical re-interview of
individuals scoring positive
FOLLOW-UP:
Clinical interview by health
worker using DIS,
BASELINE:
Self-reported using DISC-C
(administered by child
psychiatrist)
FOLLOW-UP:
Clinical interview face to face
using the SADS_L and Personality Disorder Exam (PDE)
BASELINE:
Self-reported psychosis
proneness scales:
a) Physical Anhedonia
b) Perceptual Aberration
c) Magical Ideation
d) Impulsive Nonconformity
Assesment type/ Instruments:
Baseline Exposure
follow-up Outcome
Poulton et al.,
2000
Of 7800 students, 534 subjects
were selected for 10-year
follow-up; n=508 (95%; 355
exposed, 153 non-exposed)
had outcome assessment
Follow-up clinical nonpsychotic outcome
10-year follow-up of undergraduate students with low
and high scores on four
schizotypy scales
Follow-up clinical psychotic
outcome
Baseline PSYCHOTIC predictor
(exposure)
Chapman et
al., 1994
Mean age baseline and follow
UP (Years)
Study type and goals
Study ID
Sample size (N)
Qualitative description of longitudinal studies in representative samples studying predictive value of subclinical psychotic experiences (exposure) for transition to clinical psychotic and non psychotic disorders (outcome)
Tabel 1A
129
B) EDSP: risk set differs from
Dominguez et al (2009) as not
restricted to youngest cohort;
risk sets were n=1876, n=1608
and n=1221 for bipolar
disorder, depression and
anxiety, respectively
Additional data analysis of
A) Hanssen et al (2005)
NEMESIS study
B) Dominguez et al (2009)
EDSP study:
Current report
4914 community subjects
(subjects with psychotic
disorder were screened and
excluded)
Risk set for analysis: 4726
B) EDSP: see Dominguez et al
(2009)
NEMESIS: see Hanssen et al
(2005)
29-52
B) EDSP: differs from
Dominguez et al (2009):
exposure was presence of any
T2 CIDI psychotic symptom
NEMESIS: see Hanssen et al
(2005)
Self reported psychotic
experiences:
0=no symptom
1= weak symptom (rarely /
sometimes)
2=strong symptom (often /
very often)
Degree of persistence over the
5 years from T0-T2 of selfreported SCL90-R psychotic/paranoid symptoms in
the highest 10% of scores:
once, twice or thrice.
Self-reported Youth Self Report
item age 14: ”I hear sounds or
voices that other people think
aren't there” – rarely/never vs.
sometimes/often
---
---
ICD 10 register
Non-affective psychotic
disorder
T3 Diagnosis of clinical
psychotic disorder based on (i)
help-seeking, (ii) service use
and (iii) impairment
Caseness: CIDI DSM-IV nonaffective psychotic disorder
or past medical diagnosis of
schizophrenia
EDSP: T3
1) Bipolar d.o.
2) Depression
3) Anxiety
disorder
DSM-IIIR/IV:
NEMESIS: T2
1) Bipolar d.o.
2) Depression
3) Anxiety d.o.
ICD 10 register diagnosis:
Non-psychotic disorder
(see below)
---
EDSP: see Dominguez et al.
(2009)
NEMESIS: see Hanssen et al.
(2005)
FOLLOW-UP:
National case-register clinical
hospital diagnosis
BASELINE:
Clinical interview (mental
health worker) using the PERI
(Hebrew version); a subsample
was interviewed by a psychiatrist using the SADS
FOLLOW-UP:
DIA-X/M-CIDI administered as
clinical interview by clinical
psychologists
BASELINE:
Self report questionnaires
(SCL90-R) at T0, T1 and T2
FOLLOW-UP
CIDI + self-reported past
diagnosis of schizophrenia
BASELINE
Youth Self Report hallucination
item
= T0 baseline; T1 first follow-up, T2 second follow-up, T3 third follow-up; SADS: Schedule for Affective Disorders and Schizophrenia-Lifetime version (SADS-L) (Endicott & Spitzer, 1978)DISC: Diagnostic Interview Schedule for Children (Costello, Edelbrock, Kalas, Kessler & Klaric, 1982; NIHM DIS for Children: Child Version); DIS: Diagnostic Interview Schedule for DSM-IV (Robins, Cottler, Bucholz & Compton, 1995); CIDI: Composite International Diagnostic Interview (WHO, 1992); YSR: Youth Self
Report questionnaire (Achenbach T, 1991); SCL-90-R: Self Report Symptom Checklist-90-R (Derogatis & Cleary, 1997); DIA-X/M-CIDI: updated version of the World Health Organization’s CIDI version (Wittchen & Pfister, 1997; DIA-X-Versions of the WHO
CIDI).; PERI: Psychiatric Epidemiology and Research Interview (Shrout, Dohrenwend & Levav, 1986 Hebrew version: Roberts & Vernon, 1981); BPRS: Brief Psychiatric Rating Scale (Overall & Gorham, 1962); CAN: Camberwell Assessment of Need (Slade,
Phelan, Thornicroft & Parkman, 1996)
a
NEMESIS : see Hanssen et al
(2005)
24-year follow-up through
national case register of
general population cohort
assessed for presence of
psychotic symptoms at
baseline
Werbeloff et
al.,
2009
14-21 years
Analyses restricted to youngest 14-24 years
group aged 14-17 at baseline
(T0: n= 1395, T1: n= 1228, T2:
n=1169, T3: n= 1022 [73%]).
Risk set for analysis: n=845
10-year longitudinal general
population cohort study,
assessing mental health four
times over a period of 10
years. Clinical outcomes were
assessed over a 5-year period
(from T2 to T3)
Dominguez et
al.,
2009
Birth: n=7223; age 14 years:
n=5172(72%); age 21 years:
n=3801(53 %). Risk set: 3573
Birth cohort (1981-1983)
assessed after 5 and 14 years
for psychotic symptoms and
after 21 years for screenpositive non-affective psychotic disorder (SP-NAP)
Welham et al.,
2008
Description of the possible modifiers as causes of heterogeneity of the studies
Four studies18, 19, 31, 32 used similar methodology for exposure and outcome assessment based
on the Composite International Diagnostic Interview34 whereas other studies27, 33 used different instruments. Similarly, all studies reported exposure assessment based on prevalence
estimates whereas one study also reported assessment of incident exposure18. Another important factor was that one study provided outcomes based on hospital admission33,
whereas all other studies were independent of health care use. Thus, three modifiers were
included in meta-regression analyses (psychotic/non-psychotic disorder, hospital admission
yes/no and incident/prevalent exposure); in addition sensitivity analyses excluded results
that were outliers with respect to outcome assessment (hospital admission in the study by
Werbeloff and colleagues33).
Results for dichotomous exposure meta-analysis
Original study results
Findings from individual studies are summarised in tables 1B (psychotic outcomes) and 1C
(non-psychotic outcomes).
130
131
SP-NAP
Psychotic Disorder
Hospitalization for non affective
psychosis
Welham et al., 2008
Dominguez et al.,
2009
Werbeloff et al.,
2010d
12
Strong symptoms
60
19
34
45
10
T1 Incident single PE
T1 Incident multiple PE
T1 Incident PE with depression
T1 Incident PE without depression
T1 Incident multiple PE with depression
299
463
283
213
T0 prevalent multiple PE
T0 prevalent PE with depression
T0 prevalent PE without depression
T0 prevalent multiple with depression PE
2614
652
Strong symptoms
14
Level 3
Weak symptoms
33
Level 2
---
132
Level 1
No symptoms
---
Level 0
451
447
T0 prevalent single PE
Self-reported auditory hallucinations
746
T0 Prevalent PE
T0 No prevalent PE
79
T1 Incident PE
T1 No incident PE
95
8 (1.3%)
13 (0.5%)
---
3 (27.3 %)
4 (16.0 %)
6 (5.3 %)
---
18 (4.0%)
21 (9.9%)
4 (1.4%)
22 (4.8%)
24 (8.0%)
2 (0.5%)
26 (3.5%)
4 (40.0%)
1 (2.2%)
5 (14.7%)
4 (21.1%)
2 (3.3%)
6 (7.6%)
3 (25%)
9 (9.5%)
---
1 (0.3%)
3 (0.9%)
1 (0.3%)
3 (0.9%)
4 (1.1%)
Riska exposed
---
---
1306
---
---
---
666
3112
4045
3964
---
---
654
153
153
153
153
153
N non-exposed
---
---
2 (0.1%)
---
---
---
23 (3.7 %)
38 (1.2%)
7 (0.2%)
5 (0.1%)
---
---
13 (2%)
1 (0.3%)
0 (0%)
0 (0%)
0 (0%)
1 (0.3%)
Risk non-exposed (n,%)
9.5 (1.9, 47.2)
3.6 (0.8, 16.7)
1c
9.9 (2.5-39.8)
5.0 (1.6-15.9)
1.5 (0.6 – 3.7)
1.0c
63.1 (26.5, 150.2)
8.3 (2.4, 28.4)
28.8 (12.2, 67.8)
50.3 (21.5, 117.9)
2.6 (0.5, 12.5)
20.8 (9.0, 48.1)
1c
527.9 (113.2, 2460.9)
18.0 (2.1, 157.2)
136.5 (37.4, 497.1)
211.2 (51.6, 864.1)
27.3 (5.2, 143.6)
65.1 (19.4, 218.1)
1c
16.4 (3.9, 67.8)
5.1 (1.7, 18.3)
1c
Reported relative riske
a=transition rate in the group with psychotic experiences; b= The OR’s remained virtually unchanged after controlling for sex, social class origins and age 11 IQ scores.c= Reference category, OR= Odds Ratios; d=weighted results; e= mostly adjusted ORs
from published studies, some ORs for NEMESIS and EDSP studies were calculated on the basis of new analyses for the current paper; NEMESIS study: Netherlands Mental Health Survey and Incidence Study, a longitudinal general population study. EDSP
study = Early Developmental Stages of Psychopathology Study; SPE = Subclinical Psychotic Episode; Type instruments: SADS: Schedule for Affective Disorders and Schizophrenia-Lifetime version (SADS-L). DISC: Diagnostic Interview Schedule for Children.
DIS: Diagnostic Interview Schedule for DSM-IV. CIDI: Composite International Diagnostic Interview. CBCL: Childhood Behavior Checklist. YSR: Youth Self Report questionnaire. SCL-90-R: Self Report Symptom Checklist-90-R. DIA-X/M-CIDI: updated version
of the World Health Organization’s CIDI version. PERI: Psychiatric Epidemiology and Research Interview ; Outcome: SCZ = Schizophrenia, SCF = Schizophreniform disorder, DD = Delusional Disorder, Brief PD = Brief Psychotic Disorder, Psychosis NOS =
Psychosis not otherwise specified, Psychotic BPD = Psychotic Bipolar Disorder, Psychotic MDD = Psychotic Major Depressive Disorder.
Clinical psychotic disorder
Hanssen et al, 2005
---
Weak symptoms
355
Psychotic MDD
No symptoms
355
Psychotic BPD
Poulton et al., 2000b Schizophreniform
disorder
355
355
355
N exposed
Delusional disorder
High schizotypy
Psychosis NOS
Type exposure
Type outcome
Longitudinal studies in representative samples studying predictive value of subclinical psychotic experiences (exposure) for transition to psychotic disorders (outcome): qualitative data
Chapman et al., 1994 Schizophrenia
Study
Table 1B
132
Anxiety disorder
Mania
Major depression
Anxiety disorder
1 (1.1%)
0 (0.0%)
≥ 1 psychotic symptom
lifetime
≥ 1 psychotic symptom
lifetime
219
383
304
417
695
491
22 (10.1%)
4 (1.0 %)
28 (9.2%)
14 (3.4%)
6 (0.9%)
33 (6.7%)
1002
1493
1304
3169
3972
3374
---
---
7 (1.0%)
2616
650
Strong symptoms
654
-----
654
---
654
-----
153
153
153
N non-exposed
Weak symptoms
16 (0.6%)
--32 (33.7%)
4 (33.3%)
--
--19 (20.0%)
1 (8.3%)
15 (4.2%)
7 (2.0%)
103 (29%)
Risk exposeda (n, %)
1308
--95
12
--95
12
--95
12
355
355
355
N exposed
No symptoms
No symptoms (n=654)
Weak symptoms (n=95)
Strong symptoms (n=12)
High schizotypy
Type exposure
---
76 (7.6%)
0 (0%)
79 (6.0%)
58 (1.8%)
6 (0.2%)
110 (3.3%)
---
4 (0.3%)
144 (22.0%)
-----
13 (2.0%)
---
99 (15.2%)
-----
2 (1.3%)
0 (0%)
31 (20.3%)
Risk non-exposed (n,%)
1.4 (0.8-2.2)
∞
1.6 (1.01-2.5)
1.9 (1.03-3.4)
5.8 (1.9-17.9)
2.1 (1.4-3.2)
2.1 (0.6, 7.5)
1.6 (0.5, 4.7)
1b
---
---
---
---
---
---
Reported relative riskc
= transition rate in the group with psychotic experiences b= Reference category; c= mostly adjusted ORs from published studies, some ORs for NEMESIS and EDSP studies were calculated on the basis of new analyses for the current paper.
OR= Odds Ratios. NEMESIS study: Netherlands Mental Health Survey and Incidence Study, a longitudinal general population study. EDSP study = Early Developmental Stages of Psychopathology Study.
∞ OR infinite as no transi on in non-exposed
# weighted results
a
EDSP current analysis
Major depression
NEMESIS current analysis
Mania
Hospitalization for non
psychotic disorder
Anxiety disorder
Mania
Major depression
Hypomania
Werbeloff et al., 2010#
Poulton et al., 2000
Major depression
Mania
Type outcome
Chapman et al., 1994
Longitudinal studies in representative samples studying predictive value of subclinical psychotic experiences (exposure) for transition to non-psychotic disorders (outcome):
quantitative data
Study
Table 1C
All studies showed that subclinical psychotic experiences strongly predicted clinical psychotic
outcomes. The 3-24 year risk for the exposed was in the range of 5%-25%, substantially higher than the corresponding risk in the non-exposed (ranging from 0.1-3.7%), with 3-24 year
odds ratios in excess of 10 for the strongest level of exposure (Table 1B). A degree of outcome specificity was clearly present, as ORs for non-psychotic outcomes were mostly in the
order of 2 (Table 1C).
Meta-analysis
In order to facilitate comparison, rates for all studies were uniformly transformed to express
incidence of psychotic and non-psychotic outcomes per 100,000 person-years (Fig. 1a and
Fig. 1b). This confirmed the pattern of results in Tables 1B and 1C, in that the incidence of
psychotic clinical outcome in the exposed was much higher than in the non-exposed, and
that the difference in incidence between exposed and non-exposed was much greater for
psychotic than for non-psychotic clinical outcome. In addition, the results showed that the
absolute risk for clinical outcome in the only study based on hospital admission33 was only a
fraction of the risk in studies that did not depend on service use.
be
lo
ff
et
al
et
al
al
.
an
W
er
pm
C
ha
in
gu
e
W
el
ha
m
z
et
et
al
al
et
D
om
Po
H
an
ss
ul
to
en
n
et
al
0
rate per 100,000 person years
1,000 2,000
3,000 4,000
Rate per 100,000 person years of psychotic (a) and non-psychotic (b) outcomes in exposed
and unexposed subjects (prevalent exposure only).
psychotic outcome
non-exposed
exposed
l
ta
al
W
er
be
lo
ff
e
et
C
ha
pm
an
et
gu
ez
om
in
D
H
an
lto
ss
en
n
et
et
a
al
l
al
0
rate per 100,000 person years
1,000 2,000 3,000 4,000
(a)
Po
u
Figure 1
non-psychotic outcome
non-exposed
exposed
(b)
133
Meta-analysis results for psychotic (Fig. 2a) and non-psychotic (Fig. 2b) clinical outcomes (for
the Hanssen study, based on NEMESIS data18, results with prevalence exposure were included), show that the combined yearly incidence rate of psychotic clinical outcome given
the presence of a prevalent subclinical psychotic experience was 159 per 100,000 personyears (0.2% per year) in the non-exposed, and 558 per 100,000 person-years (0.6% per year)
in the exposed. For the non-psychotic outcomes, yearly transition incidence rates for exposed and non-exposed were 1.8% and 2.6%, respectively. For non-psychotic outcomes,
confidence intervals were wide and for both psychotic and non-psychotic outcomes, confidence intervals overlap (Fig. 2a and Fig. 2b), indicating that the difference in yearly incidence
rate between exposed and non-exposed was non-significant for both psychotic and nonpsychotic clinical outcome. Heterogeneity was large (psychotic outcome non-exposed:
χ2=81.6, df=5, p<0.001; psychotic outcomes exposed: χ2=54.1, df=5, p<0.001; non psychotic
outcomes non-exposed: χ2 = 666.9, df=4, p<0.001; non psychotic outcomes non-exposed: χ2=
235.4, df=4, p<0.001).
Figure 2
Forest plot of rates per 100,000 person years in each study for the exposed and the nonexposed of psychotic (a) and non-psychotic (b) outcomes, prevalent exposure only.
Study
%
ID
ES (95% CI)
Weight
Poulton et al
198.74 (115.77, 281.70)
12.60
Hanssen et al
57.73 (14.98, 100.49)
14.50
Dominguez et al
547.45 (361.18, 733.71)
7.32
Welham et al.
175.51 (119.76, 231.27)
13.97
Chapman et al
131.58 (-50.66, 313.81)
7.49
Werbeloff et al
17.28 (8.54, 26.02)
15.30
Subtotal (I-squared = 93.9%, p = 0.000)
158.67 (69.85, 247.49)
71.18
Poulton et al
1904.76 (-230.01, 4039.54)
0.11
Hanssen et al
1182.36 (730.58, 1634.14)
2.06
Dominguez et al
3947.37 (-430.37, 8325.11)
0.03
Welham et al.
581.77 (313.79, 849.75)
4.70
Chapman et al
343.84 (149.63, 538.05)
7.00
Werbeloff et al
39.91 (10.35, 69.48)
14.93
Subtotal (I-squared = 90.8%, p = 0.000)
558.19 (170.58, 945.80)
28.82
195.25 (125.53, 264.96)
100.00
non-exposed
.
exposed
.
Overall (I-squared = 92.2%, p = 0.000)
NOTE: Weights are from random effects analysis
-8325
(a)
134
0
8325
Study
%
ID
ES (95% CI)
Weight
Poulton et al
2924.98 (2603.13, 3246.82)
12.67
Hanssen et al
1492.92 (1272.76, 1713.08)
13.70
Dominguez et al
2190.04 (1849.06, 2531.02)
12.45
Chapman et al
2417.58 (1602.77, 3232.40)
7.17
Werbeloff et al
26.47 (15.65, 37.28)
14.76
Subtotal (I-squared = 99.4%, p = 0.000)
1797.61 (500.58, 3094.64)
60.76
Poulton et al
3508.77 (487.69, 6529.85)
0.95
Hanssen et al
2642.73 (1940.72, 3344.75)
8.27
Dominguez et al
3033.71 (2236.93, 3830.48)
7.34
Chapman et al
4273.50 (3540.52, 5006.49)
7.95
Werbeloff et al
62.56 (25.60, 99.52)
14.73
Subtotal (I-squared = 98.3%, p = 0.000)
2640.92 (515.76, 4766.09)
39.24
1848.90 (1544.49, 2153.31)
100.00
non-exposed
.
exposed
.
Overall (I-squared = 99.0%, p = 0.000)
NOTE: Weights are from random effects analysis
-6530
0
6530
(b)
Meta-regression, including the four studies that used similar CIDI-based methodology for
exposure and outcome assessment18, 19, 31, 32 suggested that the effect of subclinical psychotic
experiences on psychotic clinical outcome was significant (difference in incidence between
exposed and non-exposed 648 per 100,000 person-years, p=0.043), whereas the effect for
non-psychotic clinical outcome, although directionally similar, was not statistically significant
(difference in incidence 694 per 100,000 years, p=0.31)(Table 2). Meta-regression using permutations showed similar or more conservative p-values (non-psychotic outcomes exposure
p=0.52; psychotic outcomes exposure p=0.11; T-values were all in the same direction).
Table 2
Meta-regression results stratified by psychotic/non-psychotic outcome including the four
studies that used similar methodology for exposure and outcome assessment based on the
18, 19, 31, 32
.
Composite International Diagnostic Interview
non-psychotic outcomes (6 records)
regression
coefficient
Exposure* 694
psychotic outcomes (8 records)
95% CI
p-value
regression
coefficient
95% CI
p-value
-953; 2342
0.31
648
28; 1269
0.043
*Exposure was the measure of subclinical psychotic experiences used (see Table 1)
Modifier effects
Including all studies, all psychotic and non-psychotic outcomes and all identified modifiers in
a meta-regression analysis (Table 3) revealed that (i) rates were significantly lower for psychotic outcomes, (ii) rates were significantly higher in the exposed group, (iii) incident expo-
135
sure was not a modifier and (iv) the one study analysing hospital admissions reported significantly lower rates.
Table 3
Meta-regression results, including all categories of all modifiers in one meta-regression
model.
(24 records)
Exposure*
Psychotic
Hospital admission
Incident exposure*
regression coefficient
-1706
771
-1590
120
95% CI
-2411; -1001
73; 1470
-2440; -740
-1361; 1602
p-value
0.000
0.03
0.001
0.87
*Exposure was the measure of subclinical psychotic experiences used (see Table 1)
Results for dose-response meta-analysis
Original study results
Where examined, studies reported clear dose-response relationships for variably defined
levels of exposure severity (certainty of symptom, frequency of symptom, number of symptoms, persistence over time, comorbid depression) in relation to risk of transition to psychotic clinical outcome (Tables 1A and 1B). Only weak evidence for dose-response was present for non-psychotic clinical outcome (Tables 1A and 1C).
Meta-analysis
Transformation of all studies to the same person-year denominator showed comparable
dose-response effects for psychotic clinical outcome, also for the study reporting hospitalbased outcomes33 and incident exposure assessment18. Thus, bars in Fig. 3 show that rates
increase when exposure severity increases. Meta-regression including the three studies with
linear multiple categories of exposure19, 32 18 showed statistically a significant linear increase
in yearly incidence per unit increase in exposure severity (β=962, p=0.002).
Discussion
Subjects with a history of subclinical psychotic experiences displayed higher yearly rates of
psychotic clinical outcome which was largely specific as the evidence for an increase in nonpsychotic clinical outcome was inconclusive. There was evidence for dose-response associated with number, certainty, frequency, persistence and level of affective comorbidity of
psychotic experiences.
Subclinical psychotic experiences as psychometric expression of risk for psychotic disorder
In combination, these studies provide strong evidence for the validity of the notion that subclinical psychotic experiences represent psychometric risk for later psychotic clinical outcome. Although it could be argued that CIDI measures of clinical outcome yield high rates of
false-positives, the predictive value of subclinical psychotic experiences was also apparent in
136
predicting the “hard” outcome of hospital admission. Additional validity is suggested by the
presence of dose-response and specificity.
Comparison with ’high risk’ in the ultra high-risk literature
As mentioned earlier, the two more recent systematic European and US studies on yearly
rate of transition in help-seeking individuals with ‘high risk’ psychotic experiences is around
10% 9, 10. As for the great majority of individuals in these studies high risk criteria were fulfilled because of the presence of the same “attenuated” or “transitory” psychotic experiences as reviewed here, the discrepancy between the yearly risk of 0.6% in the populationbased studies reviewed here and the 10% in the high-risk studies presented in the ultra highrisk literature requires an explanation. Even if we assume that the yearly risk of conversion in
the representative population-based studies may be as high as 1%, allowing for the possible
underestimation associated with non-linear transition patterns over time (see below), and
focussing on the most severe level of psychotic experiences, the at least 10 times higher
yearly conversion rate in help-seeking individuals presented in the ultra high-risk literature is
remarkable. An important factor that at least in part explains the discrepancy is the unspecified sample enrichment strategies that have been employed in the ultra high-risk studies.
These studies show that by conducting extensive community awareness campaigns, it is
possible to select help-seeking individuals with psychometric risk in such a way that a 10%
yearly conversion rate is brought about. The fact that many studies now have shown that this
is possible represents an important milestone for early intervention, and may benefit many
individuals at high risk for psychotic illness. The problem, however, is that the high yearly risk
of conversion observed in the ultra high-risk literature is invariably attributed to variably
defined high risk criteria, the main one being presence of “attenuated” or “transitory” psychotic experiences, whereas the current meta-analysis demonstrates that the contribution of
psychotic experiences per se is quite low. Therefore, what can be deduced is that the critical
factor feeding prediction of conversion is the use of hitherto largely unspecified sample enrichment strategies through community awareness campaigns and other means of selective
alteration of the permeability of the filters on the pathway to mental health care. The current results suggest that these sample enrichment strategies, rather than clinical high risk
criteria per se, are the critical factor allowing for the successful completion of the first stage
of any high-risk early intervention initiative: to generate a sampling frame at high enough risk
to consider the use of intervention in the first place. Therefore, there is an urgent need to
clearly describe and operationalise these strategies.
The fact that psychotic experiences were relatively specific in predicting psychotic outcomes is disappointing from the point of view of intervention at the stage of risk, as the
‘yield’ of such interventions is logically much greater if risk is associated with a much broader
outcome than clinical psychosis alone.
Relevance for the DSM-5 Risk Syndrome
For the reasons outlined above, the current results are relevant for the debate regarding the
possible inclusion of a Psychosis Risk Syndrome (PRS) in DSM-5. If the risk function in PRS is
occasioned not by psychometric criteria but instead is largely fed by unspecified sample en-
137
richment strategies that are not mentioned in the criteria, the inclusion of PRS in DSM-5
would be misleading and likely confuse its users12. Therefore, if PRS is to be included in DSM5, operationalised criteria for sample enrichment strategies need to be added. However,
given the paucity of data on replicable strategies to occasion sample enrichment, this may be
premature.
Methodological issues
In calculating yearly incidence rates, the assumption was that the rate of transition would be
spread evenly over the follow-up periods. This may not be valid, as there is some evidence
that transition rates may be higher in the first 5-10 years33. Therefore, yearly incidence rates
in two studies with longer follow-ups31, 32 may vary and be somewhat higher in the earlier
phases of the follow-up. Similarly, rates may vary according to age and sex, factors that could
not be taken into account. Nevertheless, most studies were carried out in young people in
the similar age range as samples in ultra-high risk studies.
The study by Chapman and colleagues27 was carried out in a sample of students; students are easy to recruit, but deviate from the general population as they are of younger age
and higher educational level, associated with higher and lower level of psychotic experiences,
respectively 35. Inferences drawn from student data may therefore not be generalisable to
the general population. Nevertheless, the pattern of results in the study with the student
population was very similar to that of the others, and excluding the study in the metaregression showed similar results.
Because the present analysis included only six studies and the analyses focussed on rates
rather than odds ratios, funnel plots are difficult to interpret. Publication bias cannot be
ruled out. However, given the rarity of these types of studies and the fact that included studies cannot be considered as small yet with large effect size, publication bias is unlikely.
Permutation methods within meta-regression have been developed and implemented in
Stata 26. Permutation is necessary because meta-regression gives increased rates of falsepositives when the number of studies is small and when heterogeneity is present36. In addition, in the present study the outcomes, which are rates, are not normally distributed, another reason to conduct permutations. As expected, p-values were more conservative after
permutations, but because results were similar original coefficients and p-values were presented.
Acknowledgements
This project was supported by the Netherlands Organisation for Scientific Research (NWO)
under project number: 017.002.048.
138
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
Marshall M, Lewis S, Lockwood A, Drake R, Jones P, Croudace T. Association between duration of untreated
psychosis and outcome in cohorts of first-episode patients: a systematic review. Arch Gen Psychiatry Sep
2005;62(9):975-983.
Perkins DO, Gu H, Boteva K, Lieberman JA. Relationship between duration of untreated psychosis and outcome
in first-episode schizophrenia: a critical review and meta-analysis. Am J Psychiatry Oct 2005;162(10):1785-1804.
Farooq S, Large M, Nielssen O, Waheed W. The relationship between the duration of untreated psychosis and
outcome in low-and-middle income countries: A systematic review and meta analysis. Schizophr Res
2009;109(1-3):15-23.
Marshall M, Rathbone J. Early intervention for psychosis. Cochrane Database Syst Rev 2006(4):CD004718.
Yung AR, McGorry PD. The prodromal phase of first-episode psychosis: past and current conceptualizations.
Schizophr Bull 1996;22(2):353-370.
Van Os J, Delespaul P. Toward a world consensus on prevention of schizophrenia. Dialogues Clin Neurosci
2005;7(1):53-67.
Bosanac P, Patton GC, Castle DJ. Early intervention in psychotic disorders: faith before facts? Psychol Med
Mar;40(3):353-358.
Olsen KA, Rosenbaum B. Prospective investigations of the prodromal state of schizophrenia: review of studies.
Acta Psychiatr Scand Apr 2006;113(4):247-272.
Cannon TD, Cadenhead K, Cornblatt B, et al. Prediction of psychosis in youth at high clinical risk: a multisite
longitudinal study in North America. Arch Gen Psychiatry Jan 2008;65(1):28-37.
Ruhrmann S, Schultze-Lutter F, Salokangas RK, et al. Prediction of psychosis in adolescents and young adults at
high risk: results from the prospective European prediction of psychosis study. Arch Gen Psychiatry
Mar;67(3):241-251.
Olsen KA, Rosenbaum B. Prospective investigations of the prodromal state of schizophrenia: assessment instruments. Acta Psychiatr Scand Apr 2006;113(4):273-282.
Kaymaz N, Van Os J. DSM-5 and The ‘Psychosis Risk Syndrome’: Babylonic Confusion. Psychosis 2010;DOI:
10.1080/17522431003753233.
Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies in epidemiology: a proposal for
reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA Apr 19
2000;283(15):2008-2012.
Bijl RV, van Zessen G, Ravelli A, de Rijk C, Langendoen Y. The Netherlands Mental Health Survey and Incidence
Study (NEMESIS): objectives and design. Soc Psychiatry Psychiatr Epidemiol 1998;33(12):581-586.
Bijl RV, Ravelli A, van Zessen G. Prevalence of psychiatric disorder in the general population: results of the
Netherlands Mental Health Survey and Incidence Study (NEMESIS). Soc Psychiatry Psychiatr Epidemiol
1998;33(12):587-595.
Wittchen HU, Perkonigg A, Lachner G, Nelson CB. Early developmental stages of psychopathology study (EDSP):
objectives and design. Eur Addict Res Mar 1998;4(1-2):18-27.
Lieb R, Isensee B, von Sydow K, Wittchen HU. The Early Developmental Stages of Psychopathology Study (EDSP):
a methodological update. Eur Addict Res Dec 2000;6(4):170-182.
Hanssen M, Bak M, Bijl R, Vollebergh W, van Os J. The incidence and outcome of subclinical psychotic experiences in the general population. Br J Clin Psychol Jun 2005;44(Pt 2):181-191.
Dominguez MD, Wichers M, Lieb R, Wittchen HU, van Os J. Evidence That Onset of Clinical Psychosis Is an Outcome of Progressively More Persistent Subclinical Psychotic Experiences: An 8-Year Cohort Study. Schizophr Bull
2011;37(1):84-93
Van Os J, Hanssen M, Bijl R, Ravelli A. Straus (1969) revisited: A psychosis continuum in the general population?
Schizophr Res 2000;45(1-2):11-20.
Regeer EJ, Krabbendam L, de Graaf R, ten Have M, Nolen WA, van Os J. A prospective study of the transition
rates of subthreshold (hypo)mania and depression in the general population. Psychol Med May 2006;36(5):619627.
Regeer EJ, Krabbendam L, De Graaf R, Have MT, Nolen WA, Van Os J. Berkson's bias and the mood dimensions
of bipolar disorder. Int J Methods Psychiatr Res Dec 2009;18(4):279-286.
Tijssen MJ, van Os J, Wittchen HU, Lieb R, Beesdo K, Mengelers R, Krabbendam L, Wichers M. Evidence that
bipolar disorder is the poor outcome fraction of a common developmental phenotype: an 8-year cohort study in
young people. Psychol Med Feb 2010;40(2):289-299.
139
24. Tijssen MJ, van Os J, Wittchen HU, Lieb R, Beesdo K, Mengelers R, Wichers M. Prediction of transition from
common adolescent bipolar experiences to bipolar disorder: 10-year study. Br J Psychiatry Feb 2010;196(2):102108.
25. Zimmermann P, Wittchen HU, Hofler M, Pfister H, Kessler RC, Lieb R. Primary anxiety disorders and the development of subsequent alcohol use disorders: a 4-year community study of adolescents and young adults. Psychol Med Oct 2003;33(7):1211-1222.
26. StataCorp. STATA Statistical Software: Release 11. Texas: College Station; 2009.
27. Chapman LJ, Chapman JP, Kwapil TR, Eckblad M, Zinser MC. Putatively psychosis-prone subjects 10 years later. J
Abnorm Psychol 1994;103(2):171-183.
28. Saha S, Chant D, McGrath J. Meta-analyses of the incidence and prevalence of schizophrenia: conceptual and
methodological issues. Int J Methods Psychiatr Res 2008;17(1):55-61.
29. Linscott RJ, van Os J. Systematic reviews of categorical versus continuum models in psychosis: evidence for
discontinuous subpopulations underlying a psychometric continuum. Implications for DSM-V, DSM-VI, and DSMVII. Annu Rev Clin Psychol Apr 27;6:391-419.
30. Kwapil TR, Miller MB, Zinser MC, Chapman J, Chapman LJ. Magical ideation and social anhedonia as predictors
of psychosis proneness: a partial replication. J Abnorm Psychol 1997;106(3):491-495.
31. Welham J, Scott J, Williams G, Najman J, Bor W, O'Callaghan M, McGrath J. Emotional and behavioural antecedents of young adults who screen positive for non-affective psychosis: a 21-year birth cohort study. Psychol
Med Jul 8 2008:1-10.
32. Poulton R, Caspi A, Moffitt TE, Cannon M, Murray R, Harrington H. Children's self-reported psychotic symptoms
and adult schizophreniform disorder: A 15-year longitudinal study. Arch Gen Psychiatry 2000;57(11):1053-1058.
33. Werbeloff N, Drukker M, Dohrenwend BP, Levav I, Yoffe R, Van Os J, Davidson M, Weiser M. Self-Reported
Psychotic Symptoms in the Community are Associated with Increased Risk of Later Hospitalization for NonAffective Psychotic Disorders (Conference Abstract). Schizophr Bull 2009;35(Supplement 1):74.
34. Robins LN, Wing J, Wittchen HU, et al. The Composite International Diagnostic Interview. An epidemiologic
Instrument suitable for use in conjunction with different diagnostic systems and in different cultures. Arch Gen
Psychiatry 1988;45(12):1069-1077.
35. Van Os J, Linscott RJ, Myin-Germeys I, Delespaul P, Krabbendam L. A systematic review and meta-analysis of the
psychosis continuum: evidence for a psychosis proneness-persistence-impairment model of psychotic disorder.
Psychol Med Feb 2009;39(2):179-195.
36. Higgins JP, Thompson SG. Controlling the risk of spurious findings from meta-regression. Stat Med Jun 15
2004;23(11):1663-1682.
140
CHAPTER 5
Summary
141
General comments
In this thesis, several studies are described concerning the developmental and affective aspects of schizophrenia and bipolar disorder, in terms of where they differ and overlap, and
how that potentially overlaps in the prodrome.
Examination of the developmental and non-developmental risks of affective and nonaffective psychotic disorders has important diagnostic implications, which led us to include
work focusing on examination of diagnostic likelihood ratios of developmental and nondevelopmental risks in relation to the diagnoses of schizophrenia and bipolar disorder. This is
discussed in Chapters 2 and 3.
The study of the developmental domains of affective and non-affective psychosis was
carried out by examining areas where they differ and overlap in genetic and environmental
factors shared by these two disorders (Chapter 3). In addition, several aspects of the prodrome, in terms of the distinction between bipolar disorder and schizophrenia, were investigated, examining the view that transition from subclinical states of psychometric risk in the
general population to psychotic disorder can be viewed from the perspective of a psychosis
continuum. In order to investigate this, aspects of the epidemiology of psychotic experiences
in the general population, which may predict transition to various clinical psychotic disorders,
was investigated (Chapter 4).
In the above-mentioned studies, data pertaining to both the NEMESIS and the EDSP data
were examined. Additionally, we examined evidence from the literature using the methodology of qualitative and quantitative review of both observational and experimental studies.
Our findings
In Chapter 2, the developmental domains of affective and non-affective psychosis – in particular schizophrenia and bipolar disorder – were investigated. A common and well accepted
position regarding the causes of psychiatric disorders, especially with schizophrenia, is the
supposition of a multifactorial component (Johns & Van Os, 2001) similar to chronic physiological disease such as diabetes or cardiovascular disease. In the case of schizophrenia and
bipolar disorder, this generally is taken to indicate that apart from the genetic component –
with reported heritabilities of around 80% for schizophrenia and bipolar disorder – environmental factors may play a role as well. However, it is unclear what impact genetic and environmental factors have in terms of attributable fraction within these highly heritable diagnostic constructs. It is also unclear to what extent genetic factors play in their contribution to
domains generally taken to reflect aberrant neurodevelopment, such as brain structural
alterations associated with these disorders. Since pathophysiological findings demonstrate
that there are brain changes in most psychiatric disorders, we wanted to investigate to what
extent there is a genetic contribution to the development of brain structures. We also wanted to see if this genetic influence is also the cause of the brain changes in schizophrenia,
because the brain changes in schizophrenia have been investigated more extensively than in
any other psychiatric disorder.
In the first article, ‘Heritability of structural brain traits: an endophenotype approach to
deconstruct schizophrenia’, a literature search on the heritability of brain structures in
142
healthy people, in people with schizophrenia and pedigrees was carried out. In this article,
evidence from literature on the heritability of brain structures in healthy monozygotic twins
was examined and the heritability rates in studies with monozygotic and dizygotic twins concordant and discordant for schizophrenia were summarised in order to be able to estimate
the contribution of genetic and environmental factors to brain structures. Examination of the
contribution of genetic and environmental factors was also investigated in pedigree studies,
in which the genetic component can be estimated more accurately as environmental factors
are assumed to play a minor role in pedigree effects. In addition, the genes that may be involved in these brain structures in the different subgroups were investigated.
The following conclusions were drawn from the studies: the brain structures that were
formed earlier in life and also the deeper structures were more genetically influenced than
structures formed later in life, which are more environmentally influenced. The findings were
similar for most of the brain structures across the different groups. However, there were also
conflicting results in different studies on the same brain structures both in healthy people
and in patients with schizophrenia. The brain changes in schizophrenia are present in about
50% of the patients that are included in these neuroimaging findings. As for the genes underlying these effects on brain structures, there are not enough studies showing evidence for
effects of specific genes on specific brain structures.
The use of brain structures as an endophenotype may be valuable in deconstructing the
genetics of brain traits in healthy people, people with schizophrenia and in pedigrees. Taking
into account the limitations of neuroimaging as well of genetic epidemiological studies, and
considering the problematic assumption that monozygotic twins share the (prenatal and
postnatal) environment to the same degree as dizygotic twins, that most studies included
rely on, the results of this review can only be described as helpful for further research in
defining which brain structures might be more genetically influenced and represent genetic
effects underlying the aetiology of schizophrenia. More research is necessary, and it is too
early to state to what extent brain structural alterations represent the genetics of schizophrenia, let alone bipolar disorder, in a meaningful way.
In the second article, ‘Murray et al. (2004) revisited: is bipolar disorder identical to
schizophrenia without developmental impairment?’, the developmental factors in the overlap or non-overlap between the two disorders, i.e. schizophrenia and bipolar disorder, were
investigated in a qualitative review. The search was carried out by examining the evidence
from the literature in the past ten years and by revisiting the view by Murray et al., who
published an article in 2004 which advocated the view that bipolar disorder was identical to
schizophrenia without, however, expression of developmental impairment.
Most of the findings by Murray et al. (2004) are still relevant, according to our findings
from the literature. An additional finding was that in the past ten years, more evidence has
arisen that genetic risk for schizophrenia is expressed in part as neurocognitive impairments,
whereas genetic risk for bipolar disorder is only mildly expressed with neurocognitive alterations. General population studies show that both disorders have phenotypes that are associated with psychometric risk states in healthy individuals and that these psychometric states
are similarly highly comorbid with each other. One other finding was that exposure to an
urban environment impacts on specific developmental associations with schizophrenia which
is not observed in bipolar disorder, although we did find in one of our analyses with the
NEMESIS data (Chapter 3, first article), that urbanicity specifically impacts on the psychotic
143
but not on the affective dimension of bipolar disorder, suggesting dimension-specific effects
across diagnostic categories.
In the third article, ‘Extended psychosis phenotype – yes: single continuum – unlikely’,
an alternative theory opposing the categorical view of psychiatric disorder, in the sense of ill
versus not ill (in contrast to the use of continuum as denoting a lack of contrast between
disorder categories within psychotic illness), e.g. the continuum theory of psychosis, was
investigated conceptually in an attempt to clarify whether a population psychosis continuum
or, perhaps better described as extended psychosis phenotype exists and, if so, how this extended psychosis phenotype is distributed fully or quasi-continuous, and what determines
whether some people experiencing psychotic experiences develop psychotic disorders later
in life whereas others do not. Until recently, research on syndromal clustering of dimensions
of psychosis was carried out almost exclusively in the population of people already attending
mental health services, based on the assumption that symptoms observed in patients with
psychotic disorder naturally did not exist outside mental health services. However, general
population studies are showing that syndromal clusters of psychotic and affective symptoms
not only exist in populations attending mental health services, but also are expressed as
extended phenotypes in the general population (Stip & Letourneau, 2009). In order to be
able to make a conceptual shift from studying clusters of psychosis dimensions in mental
health services to the general population, it may not be productive to consider populations
inside and outside the hospital as different at the level of symptoms per se, but at the level of
whether or not need for care has developed. Not every person with a psychotic symptom will
develop a clinical need, resulting in a visit to a general practitioner or to psychiatric services.
This subject is further investigated in the last article of chapter 4. We present in this article the first meta-analysis to ever attempt to estimate which people, given a certain level of
expression of the extended psychosis phenotype in the general population (i.e. not in highly
selected high-risk samples) will develop need for care, help-seeking behaviour and diagnostic
status. Data were extracted from all published longitudinal studies in non-help-seeking general populations. We conclude that there are a number of variables predicting the transition
of subclinical symptoms in general populations to clinical disorders. Not only the load of
psychotic experiences is important, but also what type of coping the person develops, and
the degree of persistence of these subclinical psychotic symptoms over time. We further
suggested a new model for research across the psychotic spectrum, suggesting that there is a
psychosis continuum, but not a single psychosis continuum.
In Chapter 3, studies concerning the affective domains of psychosis are presented. In the
first two articles, NEMESIS data were analysed, in search for evidence for overlap (or nonoverlap) between affective and non-affective psychosis. In the first article, ‘Evidence that the
urban environment specifically impacts on the psychotic but not the affective dimension of
bipolar disorder’, urbanicity as an environmental risk factor for bipolar disorder was investigated, because reports on this issue have been inconsistent, whereas high rates of psychotic
disorders have already been shown in numerous studies related to the urban environment.
Our hypothesis was that any effect of urbanicity on the bipolar phenotype would be moderated by comorbid psychotic symptoms. The cumulative incidence of bipolar and psychotic
symptoms and syndromes – assessed with the CIDI in relation to five levels of population
density of place of residence – were examined. In addition, we examined the degree of comorbidity between broadly and narrowly defined bipolar phenotypes on the one hand and
144
the dichotomous presence of broadly (17.2%) and narrowly (3.8%) defined psychotic symptoms on the other, as a function of population density of place of residence.
Higher rates of bipolar disorder, however defined, were observed in more urbanized areas, as well as a strong interaction between comorbid psychosis and level of urbanicity, indicating that with greater degree of psychotic comorbidity, a greater effect size of the urban
environment was observed. For bipolar disorder without psychosis, no effect of urbanicity
was apparent. These results suggest differential environmental causal effects on affective
and psychotic dimensions of bipolar psychopathology.
Given the fact that associations with urbanicity are thought to reflect the impact of an
environmental exposure that interacts with genetic liability to produce illness (Van Os,
2003a, 2003b, 2004), these results should be examined in light of previous work in the field
of both molecular genetics and genetic epidemiology that has shown that there is a substantial sharing of genetic risk between bipolar disorder and non-affective psychosis (Cardno et
al., 2002; Berrettini W, 2003a, 2003b). This means that when two individuals have a similar
amount of shared genetic liability for both bipolar disorder and schizophrenia, the one that
becomes exposed to an urban environment may develop a more psychotic illness, whereas
the one not exposed to an urban environment may develop a more mania-only illness, suggesting that within the bipolar spectrum the impact of urban environment on the occurrence
of more psychotic illness may be mediated by a different pathway than the rate of more
mania-only illness. These mechanisms of gene-environment interaction have also been described for other psychiatric disorders such as depression and anxiety, two conditions that
have been shown to share genes that, however, produce differential outcomes depending on
subsequent exposure to divergent environmental risk factors, some resulting in anxiety outcomes and other in depression (Kendler et al., 1987, 1992, Kendler, 1996).
In the second article, ‘The impact of subclinical psychosis on the transition from subclinical mania to bipolar disorder’, the prevalence of subclinical psychotic and manic symptoms was investigated, in terms of how these subclinical population phenotypes co-vary with
and impact on each other. Again, NEMESIS data were used and the degree of comorbidity
between subclinical mania and subclinical psychosis was investigated. In addition, the impact
of subclinical comorbidity on social impairment and on the transition from subclinical mania
to onset of bipolar disorder was also investigated.
The lifetime prevalences of at least one manic and one psychotic symptom were 4.1%
and 4.2%, respectively, but, after excluding the people with DSM-III-R diagnoses of bipolar
disorder or psychotic disorder, these prevalences were 2.3% (subclinical mania) and 2.8%
(subclinical psychosis). Regarding the question as to how these phenotypes co-vary with each
other, it was found that individuals with subclinical mania had a 17% risk of subclinical psychosis, compared to 2.3% in those without (p < 0.000). Subclinical psychosis in individuals
with subclinical mania was much more predictive of a future diagnosis of bipolar disorder. As
for social impairment, there was a positive interaction between social impairment due to
physical and psychological problems and subclinical psychosis, indicating that for a given
level of subclinical mania, the coexistence of subclinical psychotic symptoms was more predictive of social impairment, although this statistical interaction was not significant.
Thus, the subclinical phenotypes of mania and psychosis are more prevalent than their
clinical counterparts and cluster together. One of the mechanisms by which the clustering of
subclinical mania and subclinical psychosis may be relevant for clinical outcomes is that pos-
145
sibly the formation of more toxic combinations of subclinical mania and subclinical psychosis
may facilitate a higher transition to bipolar disorder. A better understanding of this pathway
is crucial for the development of early intervention.
The study is important in showing that subclinical symptoms should not be neglected,
since they result in higher transition rates from subclinical to clinical disorders; specifically
the comorbidity of psychotic symptoms may be toxic in this regard. Subclinical phenotypes
can be seen as intermediary phenotypes of a mood continuum that after exposure to additional risk factors may progress to a full-blown disorder (Hanssen et al., 2005), specifically in
the presence of psychotic symptoms at the subclinical level.
In the third article, ‘Evidence that patients with single versus recurrent depressive episodes are differentially sensitive to treatment discontinuation: a meta-analysis of placebocontrolled randomized trials’, the focus was on the maintenance treatment of depression, as
a paradigm for sensitisation. Antidepressants are effective in the treatment of depression,
but also in prevention of relapse after remission from an acute episode. It is unclear though,
to what extent the prophylactic effect of antidepressants is moderated by the duration of the
continuation phase, level of abruptness of antidepressants discontinuation, or the number of
previous episodes. This study attempted to address these questions.
All published randomized, placebo-controlled, double-blind clinical trials available before
May 2007 were identified that addressed the efficacy of continuation or maintenance treatment of major depressive disorder with either SSRIs or TCAs and included patients entering a
maintenance phase after achieving remission from the acute phase.
The conclusions were that the overall reduction of relapse risk in the maintenance phase
was highly significant for both SSRIs and TCAs over one year of follow-up of maintenance
treatment. Treatment with an antidepressant results in approximately 70% reduction of risk
of relapse, confirming previous findings by Loonen et al. (1991), Geddes et al. (2003) and
Viguera et al. (1998). The prophylactic effect appeared to be constant over the length of the
continuation phase. The major conclusion in this study was that recurrent episode patients
experience less protection from antidepressants over the maintenance phase than single
episode patients. This suggests that with increasing number of episodes, patients may develop a relative resistance against prophylactic properties of antidepressant medication.
There was no difference between abrupt discontinuation of antidepressants versus gradual
discontinuation on relapse rates, except for a particular subgroup of recurrent episode patients, for which the mode of discontinuation was important. In these patients, abrupt discontinuation may lead to a relapse, a problem that has been identified for lithium as well.
This finding may be interpreted in the context of sensitisation or a kindling-like process, in
which biochemical and physiological processes involved in the illness become progressively
more easily triggered by the same circumstances or precipitants.
We develop the hypothesis that sensitisation, as a form of progressive behavioural sensitisation, would be evident in depressive disorder manifested by evidence of decreased response to treatment with an increasing number of depressive relapses. However, the sensitisation model may not only be applicable to depression. There is accumulating evidence that
behavioural sensitisation may also be relevant for psychotic disorder and in particular for
positive symptoms in psychotic disorders (Collip et al., 2010). In other words, sensitisation
may be important for the affective pathway in psychotic disorder (Myin-Germeys & Van Os,
2007). Future studies should take this issue further, possibly focusing on both affective and
146
psychotic symptoms in psychotic disorder, also in relation to antidepressant treatment in this
group.
In Chapter 4, we described to what extent findings in the literature and research in psychiatry on developmental and affective domains of psychosis are applicable to the prodrome.
Research on the prodrome is important because it is known that disorders such as schizophrenia or bipolar disorder do not have a sudden onset. In many patients, the first episode is
preceded by a lengthy pre-phase in which symptoms manifest themselves in attenuated form
for prolonged periods, sometimes as long as years. These trajectories may be amenable to
early intervention, in order to reduce the psychological, social and possibly biological disruption that can lead to poor outcome (Pantelis et al., 2003).
Although there are many studies on the prodrome, these may be biased to a degree due
to the vagaries of the selection processes that underlie the ultra-high-risk samples. Moreover, there is considerable confusion about the meaning conveyed by the term at risk mental
state.
In the first article, ‘DSM-V and the ‘Psychosis Risk Syndrome’: Babylonic confusion’, the
focus is on whether DSM-V should consider the inclusion of a new category, called the psychosis risk syndrome – or possibly the related term attenuated psychotic symptoms syndrome. We make the point that this should not occur, since risk is not an appropriate term in
this context because subjects already, by definition, have a need for care, and the predictive
value in even ultra-high-risk samples is too low for use in clinical practice.
In the second article, ‘The case of the missing evidence: what do subclinical psychosis
spectrum experiences predict in unselected representative population samples? A systematic review enriched with new results’, a meta-analysis on psychotic experiences in the general population is presented, with a specific focus on unselected general population samples
with psychotic experiences that are at risk of making the transition to a clinical disorder. The
yearly risk of conversion to a clinical psychotic outcome in exposed individuals (0.56%) was
3.5 times higher than for individuals without psychotic experiences (0.16%). Conversion to a
clinical disorder increased with the number, certainty, frequency, persistence and degree of
affective dysregulation of psychotic experiences. The major finding in this study was the
discrepancy between the 10% conversion rate in the high-risk literature compared to the
actual 0.56% conversion rate in unselected population-based samples. The explanation for
the discrepancy lies in the sample enrichment strategies through community awareness
campaigns and other selective inclusion methods by which people participating in these highrisk studies are selected, creating a very high density of risk. The high conversion rates of the
high-risk early intervention studies thus are based on sample enrichment strategies, rather
than on clinical high-risk criteria per se.
147
CHAPTER 6
Future directions
149
Future direction of diagnosis across the spectrum of psychosis
Given the results of the studies included in this thesis, we can conclude that a multidimensional approach to the broad range of psychotic disorders, whether it is the affective domain
or non-affective psychotic (syndromal) domain, would be more useful (Van Os et al, Nature,
2010) than conceiving of these syndromes as categorical entities. Thus, psychotic disorders
may be viewed as the variable expression of a general psychotic syndrome consisting of various symptom dimensions that cross-sectionally cluster together in different degrees and
combinations in different people. These symptom dimensions may, longitudinally, differ
within the same people over time.
In this multidimensional approach four main dimensions of symptoms characterize the
general psychotic syndrome: affective dysregulation (depression, mania or anxiety), psychosis (delusions, hallucinations), negative symptoms (motivational impairment) and cognitive
alterations. In the general population, low levels of these four dimensional phenotypes represent the behavioural expression of vulnerability (extended phenotype in Figure 2, prevalence of 10%-20%) for psychosis. Instead of estimating heritability rates for the diagnostic
construct schizophrenia, from a scientific point of view it is more useful to investigate the
heritability rates of these four symptom dimensions. The four symptom dimensions display
moderately high heritability levels of their own, varying from 40% heritability rates for affective and psychosis domains to a range of 40%-60% of heritability rates for cognitive domains.
For example, in the study by Chen and colleagues (2009), it was shown that in a sample of
278 subjects consisting of patients with schizophrenia, their non-psychotic siblings, healthy
controls and their healthy full siblings, many clinical cognitive domains were impaired in
patients with schizophrenia and their non-psychotic siblings compared to healthy controls
and siblings of these controls. Negative symptoms, working memory, episodic memory and
executive function were found to be significantly heritable in sibling pairs. Significant genetic
correlations were observed between negative symptoms and the cluster of working memory,
episodic memory and executive function. Similar findings on the heritability of cognitive
functions were found by Antila and colleagues (2009). They found that the heritability of
cognitive functions was generally similar irrespective of psychopathology in families of bipolar I disorder patients (bipolar families) and unaffected relatives with bipolar I disorder only
in the family compared to another group of families with both bipolar I disorder and schizophrenia or schizoaffective disorder (mixed families).
The symptom domains, at subclinical level, show low levels of correlation in the general
population as indicated by less overlap at the level of behavioural expression of vulnerability
(heritability of 40%-60%) seen in Figure 2. The heritability of schizophrenia is around 80%90%, however these high heritability rates are not purely due to genetic factors. Part of the
high heritability rate is due to the effect of genetic and environmental interactions that give
rise to more severe levels of phenotypic expression, which in turn contribute to the likelihood of clinical detection by psychiatric services, by passing the filters on the pathway to
mental health care (see mental health care filter in Figure 2). The symptom dimensions contribute independently to the clinical detection by psychiatric services and give rise to clinical
diagnostic descriptions such as schizophrenia that have a relatively low prevalence on its
own. However, these four symptom dimensions are subject to comorbidity bias (Berkson’s
150
bias), meaning that they are much more correlated and comorbid at the level of psychiatric
services.
Current psychiatric diagnostic entities may not be as scientific or useful as we think. Ideally, a diagnosis in psychiatry should fulfil the following criteria (Van Os, 2010): the diagnostic
construct of disease has: 1) some link with a measurable phenotype in nature, 2) is valid in
combining aetiological, symptomatic, prognostic and treatment specificity, 3) is useful and 4)
is acceptable to those who are invited to carry its label. These four elements do not apply to
schizophrenia or bipolar disorder, or to any other psychiatric disorder (Bentall, 2004; Brockington, 1992; Kendell & Jablensky, 2003; van Praag, 1976). Psychosis is easy to identify, but it
is not exclusive to schizophrenia, and it occurs across a range of diagnostic categories of
psychotic disorders. And, more important, the psychotic symptoms that are known to characterize psychotic patients are also prevalent in the general population at a subclinical level
and show characteristic patterns of clustering that strongly resemble the disease phenotype.
Van Os and colleagues (2000) showed that in the general population, subclinical delusional
experiences are strongly associated with subclinical hallucinatory experiences, and that similarly, subclinical positive psychotic experiences are associated with depressive symptoms and
blunting of affect. The prevalence and incidence of positive psychotic experience is much
higher than reported rates for the (clinical) psychotic disorder (Van Os et al., 2009). The same
data exist with regard to symptoms pertaining to the bipolar spectrum (Angst, 2007).
Figure 2
Complexity of the psychotic disorder phenotype in aetiological research (Van Os et al, Nature, 2010)
This broader syndromal view of clustered dimensions, depicted in Figure 2, may be more
useful for aetiological research as the impact of environmental influences in psychotic disorders appears to be dimensional rather than categorical. The subclinical expressions of these
four symptom dimensions are also encountered in the non-diagnosed general population at
a rate of 10%-20% (Van Os & Kapur, S, 2009; Dominguez et al., 2010) and are also associated
151
with the same demographic, familial and other risk factors as the clinical syndrome of schizophrenia and related psychotic disorders (Dominguez et al., 2010; Polanczyk et al., 2010; Van
Os et al., 2009; Kelleher & Cannon, 2010). Therefore, this model is not only applicable to
those who have permeated the filters of mental health care and thus are already in need for
care, but also to those in the general population, who might show transient low levels of
these dimensional phenotypes, representing the behavioural expression of vulnerability for
the different dimensions of the psychotic range. The heritability levels of these four dimensions can be estimated taking into account the specific factors that impact this vulnerability,
but also the non-specific commonalities, which are a result of the gene-environment interplay.
Research to date has shown that the cognitive alterations, psychosis (hallucinations and
delusions) and affective dysregulations observed in psychotic disorders such as schizophrenia
and bipolar disorder, are also encountered in common mental disorders, like anxiety disorders and depression. Differences between these groups are quantitative rather than qualitative (Hanssen et al., 2003; Weiser et al., 2008). The relative non-specificity of symptoms in
psychiatric disorders extends to the level of familial aggregation (Van Os et al, 2010). For
example, siblings of patients with common mental disorders also display cognitive alterations
(Weiser et al., 2008) and brain changes (Boos et al., 2008), albeit to a lesser degree. Although
a family risk of history of schizophrenia is associated with the strongest relative risk, almost
any psychiatric disorder in first-degree relatives is associated with an increased risk for schizophrenia. This means, that in terms of attributable risk, 30% of schizophrenia in the general
population can be attributed to psychiatric family history in general, compared to 6% that is
attributable to a family history of schizophrenia specifically (Mortensen et al, 2010). This
suggests that, in addition to possible specific factors impacting on liability to specific psychiatric disorders, shared genetic and environmental factors may also impact on broader neurodevelopmental alterations resulting in liability to broad dimensions of mental health (Argyropoulus et al., 2008).
Future directions high-risk and aetiological research across the
spectrum
After decades of research, schizophrenia and related psychotic disorders are still among the
most debilitating disorders in medicine. Research on the psychosis risk syndrome (or attenuated psychotic symptoms syndrome), known variably as clinical high risk, or ultra-high risk
mental state was inspired by observations of chronic illness courses in most individuals,
greater treatment responsiveness during the first episode, possible progressive gray matter
decline during early disease stages, and retrospective accounts of prodromal or early illness
signs and symptoms. Research on psychosis risk syndrome was mainly focused on evidence
from research in high-risk groups selected for presence of need for care, and has a focus on
prevention of deterioration of the already ongoing onset of psychosis in this subgroup of
patients, who have already passed the filters on the pathway to mental health care. However, the examination of conversion rates in general population cohorts show only minimal
yearly conversion risks, throwing doubt on the high-risk paradigm. Results from intervention
152
studies are encouraging, but only apply to highly selected and non-representative samples.
Therefore, they cannot yet be generalised to general psychiatric care. The next phase of
research on psychosis risk syndrome has moved to larger, multicentre projects to increase
generalisability and to ensure that sufficiently large samples at true risk for psychosis are
included. However, samples are still highly selected and artificial density of risk of conversion
is brought about by non-replicable sample enrichment strategies. Thus, it may be misleading
that much emphasis in these new studies is on identification of biomarkers for conversion to
psychosis that mainly focus on neuroimaging and genetics because results are only valid in
the uniquely selected sample in which they are produced, which precludes generalisation.
None of these high-risk studies to date have shown replicated evidence for biological or nonbiological causal factors. However, one positive note on high-risk research conducted to date
is that it has shown that it is time to critically re-examine our diagnostic entities, the methodological approaches and consequences of sampling strategies in high-risk studies, and the
need for a priori formulated hypotheses.
Future research therefore should attempt to integrate the genetic and epidemiological
aspects of the psychosis spectrum in representative longitudinal general population samples
followed over time. Genetic epidemiology should integrate the study of the relevant liability
phenotypes with variables relating to the timing, severity and experience of environmental
exposures within the same study design. This suggests that population cohort study approaches need be combined with clinical sample approaches, in order to investigate all syndromes across the entire non-clinical and clinical spectrum. This would represent a novel
approach, as such work has not been undertaken before. Combining genome-wideassociation studies with environmental-wide assessments and exploring the geneenvironment wide-interactions (GEWIS) (Van Os & Rutten, 2009) may represent the next step
in future study designs. From these studies, information on the genetic and environmental
determinants of onset of psychosis in representative population samples, describing the
trajectory from risk state to clinical outcome, may be produced, helping us to understand the
true nature of psychosis and its disorders more thoroughly.
153
CHAPTER 7
Summary in Dutch
Nederlandse samenvatting
155
Inleiding
Deze thesis behandelt verschillende onderzoeken naar een mogelijke overlap van schizofrenie en bipolaire stoornis. We zijn uitgegaan van de klassieke diagnoses van schizofrenie
(niet-affectieve psychose) en bipolaire stoornis (affectieve psychose), zoals geclassificeerd in
de diagnostische handboeken DSM-IV en ICD-9 (Hoofdstuk 1). In de affectieve psychose staat
de stemmingscomponent op de voorgrond, terwijl in de niet-affectieve psychose de cognitieve component de hoofdrol speelt. We hebben naar een eventuele overlap, of juist een gebrek daaraan, gezocht op een tweetal domeinen van de affectieve en niet-affectieve psychose, nl. het ontwikkelings- en het affectieve domein; verder hebben we gekeken hoe deze
domeinen van invloed zijn op de prodromale fase van de twee stoornissen.
Bevindingen
In hoofdstuk 2 begonnen we met het ontwikkelingsdomein van de affectieve en nietaffectieve psychoses, in het bijzonder keken we naar de diagnoses bipolaire stoornis en schizofrenie. We deden dit aan de hand van 3 artikelen.
Een theorie met brede consensus over de oorzaken van psychiatrische aandoeningen is,
dat het ontstaan kan worden teruggevoerd op een multifactoriële aetiologie, d.w.z. dat er
meerdere factoren bijdragen aan het ontstaan van de aandoening. Schizofrenie en bipolaire
stoornis kennen een erfelijkheid van rond de 80%, maar onduidelijk is hierin de bijdrage van
gen/omgevingsinteracties. De genetische contributie als oorzakelijke factor kan tot expressie
komen in de ontwikkeling van het brein en worden onderzocht met beeldvormend onderzoek van de hersenen van patiënten met schizofrenie in vergelijking met (i) hun broers/ zusters/familieleden en (ii) gezonde controlepersonen.
Omdat de ontwikkelingsneurologische breinveranderingen het beste onderzocht zijn in
de schizofrenie, hebben we in het eerste artikel van hoofdstuk 2, Heritability of structural
brain traits: an endophenotype approach to deconstruct schizophrenia, een literatuurstudie
gedaan naar de erfelijkheid van breinstructuren in 1) gezonde personen, 2) personen met
schizofrenie en 3) niet-menselijke primaten, om zo een conclusie te kunnen trekken over de
invloed van genetische en omgevingsfactoren op breinstructuren. Daarnaast hebben we
gekeken of er genen zijn die bij de ontwikkeling van specifieke breinstructuren betrokken
zijn.
We kwamen tot de volgende conclusies:
1) De breinstructuren die vroeg in het leven ontwikkeld zijn, of die diep in het brein liggen,
ondervinden meer invloed van genetische factoren dan breinstructuren die later in het
leven gevormd zijn en meer worden beïnvloed door omgevingsfactoren.
2) In de gepubliceerde onderzoeken zijn tot op heden geen specifieke genen gevonden die
coderen voor specifieke breinstructuren; er is nog maar weinig onderzoek verricht op
dit gebied.
3) Ondanks de beperkingen van neuroimaging en genetisch onderzoek, kunnen de bevindingen tot nu toe als leidraad dienen voor verder onderzoek naar genetische oorzaken
die aan de basis van schizofrenie liggen.
156
In het tweede artikel, Murray et al. (2004) revisited: is bipolar disorder identical to schizophrenia without developmental impairment, hebben we gekeken naar een mogelijke
overlap (of juist gebrek daaraan) van ontwikkelingsfactoren bij enerzijds schizofrenie en anderzijds bipolaire stoornis. De volgende conclusies konden worden getrokken:
1) De bipolaire stoornis is in vele opzichten “identiek” aan schizofrenie, maar mist, of vertoont in veel mindere mate, de ontwikkelingsneurologische beperkingen die de schizofrenie kenmerken zoals beschreven door Murray et al, 2004.
2) Uit de onderzoeken van de afgelopen 10 jaar komt naar voren dat het genetische risico
op schizofrenie onder meer tot expressie komt als neurocognitieve beperking, terwijl
dat bij de bipolaire stoornis veel minder het geval is.
3) Urbanisatie (leven in de grote stad) als omgevingsfactor heeft invloed op het voorkomen van de ontwikkelingsneurologische stoornis schizofrenie, maar niet op het voorkomen van bipolaire stoornis.
In het derde artikel, Extended psychosis phenotype-yes: single continuum-unlikely, is de
zogenaamde continuümtheorie van psychosen onderzocht en is getracht een conceptueel
model te maken van een dergelijk psychosecontinuüm (in de zin van overlap met de normale
mentale gesteldheid in de algemene populatie; ook wel extended psychosis continuum genoemd): wat maakt dat sommige mensen over het continuüm bewegen van lage waarden
naar de hogere, klinische relevante expressiewaarden van psychose? Hiertoe werd onder
andere een meta-analyse uitgevoerd (zie hieronder).
De conclusies die getrokken werden uit dit onderzoek zijn:
1) Er zijn meerdere variabelen die de overgang van subklinische symptomen in de algemene bevolking naar een klinische stoornis kunnen verklaren.
2) Niet alleen de hoeveelheid psychotische ervaringen, en hun frequentie, zijn belangrijk,
maar ook wat voor soort mechanismen iemand ontwikkelt om hiermee om te gaan,
alsmede de mate van persistentie (duur) van de klachten over de tijd.
3) Op basis van de huidige gegevens in de literatuur valt nog niet met zekerheid te zeggen
of er werkelijk sprake is van een lineair continuüm, of dat er toch mogelijk kwalitatieve
verschillen optreden op het uiterste einde van het continuüm.
In hoofdstuk 3 zijn affectieve (stemmings-) domeinen van psychose onderzocht aan de hand
van 3 artikelen. In de eerste twee artikelen zijn data van de NEMESIS-studie geanalyseerd en
is er onderzoek gedaan naar evidentie van overlap (of gebrek aan overlap) tussen affectieve
en niet-affectieve psychoses.
In het eerste artikel, Evidence that the urban environment specifically impacts on the psychotic but not on the affective dimension of bipolar disorder, is onderzoek gedaan naar
wonen in de stad als omgevingsrisicofactor bij bipolaire stoornissen.
De conclusie van de analyses was dat urbaniteit invloed heeft op het voorkomen van bipolaire stoornis, maar alleen in samenhang met psychotische comorbiditeit: het voorkomen
van bipolaire stoornis zonder psychose is niet afhankelijk van de mate van stedelijkheid van
de omgeving. Bij gelijke genetische kwetsbaarheid kan blootstelling aan urbaniteit dus meer
schade aanrichten op het ontwikkelingsneurologisch domein.
157
In het tweede artikel, The impact of subclinical psychosis on the transition from subclinical
mania to bipolar disorder, is de prevalentie van subklinische psychotische en manische
symptomen onderzocht, met name hoe deze subklinische symptomen covariëren en op welke manier ze elkaar beïnvloeden.
De conclusies waren:
1) Subklinische manie in aanwezigheid van subklinische psychotische symptomen is meer
voorspellend voor een toekomstige diagnose van bipolaire stoornis (is dus meer
“toxisch” voor het beloop).
2) Er is een suggestieve positieve interactie tussen het optreden van sociale beperkingen
in het kader van medisch-psychiatrische aandoeningen en subklinische psychose; de
aard van de interactie was dat voor een gegeven niveau van subklinische manie, de
(co-) aanwezigheid van subklinische psychotische symptomen meer voorspellend is
voor het optreden van sociale beperkingen.
In het derde artikel, Evidence that patients with single versus recurrent depressive episodes
are differentially sensitive to treatment discontinuation: a meta-analysis of placebocontrolled randomized trials, is via de methode van de meta-analyse het fenomeen van de
progressieve sensitisatie bij affectieve dysregulatie nader onder de loep genomen. Het onderzoek richtte zich op de onderhoudsbehandeling van depressies. Antidepressiva zijn effectief in behandeling van acute depressies en in het voorkomen van terugval na een acute fase
(‘relapse’). De vraag van het onderzoek was met name of de hoeveelheid eerdere depressieve periodes de beschermende effecten van antidepressiva kunnen beïnvloeden, wat compatibel is met het idee van progressieve sensitisatie: de kans om depressief te worden naar
aanleiding van een zelfde hoeveelheid stress wordt met het verstrijken van de tijd steeds
groter omdat het individu gesensitiseerd is. De conclusie van de meta-analyse was inderdaad
dat patiënten die eerdere depressieve periodes achter de rug hebben, minder profiteren van
het profylactisch (beschermend) effect van antidepressiva dan patiënten die een eerste periode van depressie doormaken. Dit kan worden verklaard door ‘sensitisatie’ of ‘kindling-like’
processen: door de ontstane gevoeligheid worden de biochemische en fysiologische processen die tot de ziekte leiden gemakkelijker in gang gezet. Gezien het feit dat affectieve dysregulatie een kerncomponent is van psychose, verwachten we dat deze sensitisering ook een
rol speelt bij psychose.
In hoofdstuk 4 wordt aan de hand van 2 artikelen beschreven in welke mate de bevindingen
uit de literatuur over de ontwikkelings- en affectieve domeinen van psychoses, toe te passen
zijn op de prodromale fase. Onderzoek van de prodromale fase is erg belangrijk, omdat
stoornissen als schizofrenie of bipolaire stoornissen meestal niet een plotselinge start of
ontstaan kennen, maar vaak een lange prodromale fase hebben, waarin vroege interventie
mogelijk is. Er is veel onderzoek gedaan naar deze prodromale fase, maar de onderzoeken
geven mogelijk een vertekend beeld (bias) vanwege een idiosyncratische selectie van ‘ultrahigh risk’ populaties, waardoor de voorspellende waarde meer wordt bepaald door de selectieve samenstelling van de onderzochte groep dan door ‘ultra high risk’- criteria, met name
attenuated psychotic symptoms. Echter wat nu de voorspellende waarde van attenuated
psychotic symptoms in de algemene populatie is, blijft onbekend.
158
In het eerste artikel, DSM-5 and the ‘Psychosis Risk Syndrome’: Babylonic confusion, wordt
de DSM-V (in opzet) nader bekeken, met name wat betreft de eventuele inclusie van een
psychosis risk syndrome (of mogelijk de daaraan verwante term attenuated psychotic symptoms syndrome). We tonen aan dat dit mogelijk verkeerd zou zijn, met name vanwege de
boven beschreven selective sampling in de onderzoeken en het verkeerde gebruik van het
woord "risk" in psychosis risk syndrome.
In het tweede artikel, The case of the missing evidence: what do subclinical psychosis spectrum experiences predict in unselected representative population samples? A systematic
review enriched with new results, wordt een meta-analyse uitgevoerd met betrekking tot
psychotische ervaringen in niet-geselecteerde representatieve populaties uit de algemene
bevolking, en de voorspellende waarde van deze ervaringen voor het ontstaan van een psychotische stoornis. De conclusies waren dat:
1) het jaarlijkse risico op het ontwikkelen van een psychotische stoornis bij mensen met
een of meer subklinische psychotische ervaringen (0,56%) 3,5 keer hoger is dan voor
mensen zonder een psychotische ervaring,
2) het risico op transitie van subklinische psychotische ervaring naar psychotische stoornis
toeneemt met aantal, frequentie, persistentie en mate van affectieve dysregulatie van
de psychotische ervaringen,
3) de hoofdbevinding van de studie is de discrepantie tussen het 10% jaarlijkse transitierisico uit de ultra high risk- literatuur en het 0,56% transitierisico in ongeselecteerde algemene bevolkingsonderzoeken. De verklaring voor deze forse discrepantie ligt ons inziens in de manier van selective sample enrichment die wordt toegepast in de ultra-high
risk- onderzoeken maar onterecht wordt toegeschreven aan psychopathologische ‘ultra-high risk’ -criteria.
Toekomstige ontwikkelingen
Naar aanleiding van onze publicaties en literatuurgegevens kan geadviseerd worden om de
psychotische stoornis, zowel affectieve als niet-affectieve, multidimensioneel te benaderen;
dit vanwege de beperkingen die een categorale indeling van psychiatrische stoornissen met
zich meebrengt voor de dagelijkse klinische praktijk, voor wetenschappelijk onderzoek in de
psychiatrie en voor de hanteerbaarheid van deze diagnoses voor patiënten en familieleden.
Daarbij kan worden uitgegaan van verschillende symptoomdimensies, die cross-sectioneel in
verschillende mate en in verschillende combinaties aanwezig zijn in verschillende individuen,
zoals voorgesteld door Van Os en collega’s in een recente publicatie (Nature, 2010). In dit
multidimensionele model kan uitgegaan worden van in ieder geval vier symptoomdimensies
in het psychosesyndroom: affectieve dysregulatie (depressie, manie, angst), psychose (wanen, hallucinaties), negatieve symptomen (o.a. verminderde motivatie) en cognitieve veranderingen. In de algemene bevolking zijn lage gradaties van deze vier symptoomdimensies ook
aanwezig, die beschouwd kunnen worden als de expressie van genetische en niet-genetische
kwetsbaarheid voor psychose (prevalentie van rond de 10-20%). In plaats van ons vast te
bijten op de erfelijkheid van grote en vage diagnostische entiteiten zoals schizofrenie, waarbij onduidelijk is welk deel van het erfelijkheidspercentage werkelijk door genetische factoren wordt bepaald en welk deel door gen/omgevingsinteractie, is het misschien beter om de
159
erfelijkheid van deze vier symptoomdimensies apart te bepalen. De symptoomdimensies zelf
tonen matig tot hoge erfelijkheidspercentages, variërend van 40% erfelijkheid voor affectieve
dimensies tot 40-60% erfelijkheid voor cognitieve dimensies.
Toekomstige onderzoekslijnen zouden de genetische en epidemiologische aspecten van
het psychosespectrum moeten integreren; genetische epidemiologie zou het relevante fenotype (of de symptoomdimensie) moeten koppelen aan variabelen van timing, ernst en invloed van omgevingsfactoren, in een en dezelfde studieopzet. Dat betekent dat populatiecohortonderzoeken gecombineerd zouden kunnen worden met klinische onderzoeken om zo
alle syndromen van het klinische en niet-klinische spectrum bij elkaar te brengen. Op deze
manier zou het moleculair genetisch onderzoek (zoals Genome-wide association studies
[GWAS] en de speurtocht naar zeldzame varianten met grote of kleine effecten) gecombineerd kunnen worden met onderzoek naar omgevingsfactoren, en gen/omgevingsinteracties
in kaart kunnen worden gebracht.
Referenties
Angst J (2007). The bipolar spectrum. Br J Psychiatry. Mar 190, 189-191.
Antila M, Partonen T, Kieseppä T, Suvisaari J, Eerola M, Lönnqvist J, Tuulio-Henriksson A (2009). Cognitive functioning of bipolar I patients and relatives from families with or without schizophrenia or schizoaffective disorder. J
Affect Disord. Jul;116(1-2):70-9.
Argyropoulos SV, Landau S, Kalidindi S, Toulopoulou T, Castle DJ, Murray RM, Picchioni MM (2008). Twins discordant
for schizophrenia: psychopathology of the non-schizophrenic co-twins. Acta Psychiatrica Scandinavica, 118 (3),
214–219.
Bentall R (2004). Madness explained. Psychosis and human nature. London: Penguin Books.
Berrettini W (2003a). Bipolar disorder and schizophrenia: not so distant relatives? World Psychiatry, 2(2):68-72.
Berrettini W (2003b). Evidence for shared susceptibility in bipolar disorder and schizophrenia. Am J Med Genet C
Semin Med Genet, 123C(1):59-64.
Boos HB, Aleman A, Cahn W, Hulshoff Pol H & Kahn RS (2008). Brain volumes in relatives of patients with schizophrenia: a meta-analysis. Arch Gen Psychiatry, 64(3), 297-304.
Brockington, I (1992). Schizophrenia: Yesterday’s concept. Eur Psychiatry, 7, 203-207.
Cardno AG, Rijsdijk FV, Sham PC, Murray RM, McGuffin P (2002). A twin study of genetic relationships between
psychotic symptoms. Am J Psychiatry, 159(4):539-45.
Chen LS, Rice TK, Thompson PA, Barch DM, Csernansky JG (2009). Familial aggregation of clinical and neurocognitive
features in sibling pairs with and without schizophrenia. Schizophr Res. Jun;111(1-3):159-66.
Collip D, Oorschot M, Thewissen V, Van Os J, Bentall R & Myin-Germeys I (2010). Social world interactions: how
company connects to paranoia. Psychol Med, 1-11.
Dominguez, MD, Saka, MC, Lieb R, Wittchen HU & Van Os J. (2010). Early expression of negative/disorganized symptoms predicting psychotic experiences and subsequent clinical psychosis: a 10-year study. Am J Psychiatry,
167(9), 1075-1082.
Geddes JR, Carney SM, Davies C, Furukawa TA, Kupfer DJ, Frank E, Goodwin GM (2003). Relapse prevention with
antidepressant drug treatment in depressive disorders: a systematic review. Lancet, 361(9358):653-61.
Hanssen M, Peeters F, Krabbendam L, Radstake S, Verdoux H, & Van Os J (2003). How psychotic are individuals with
non-psychotic disorders? Soc Psychiatry Psychiatr Epidemiol, 38(3), 149-154.
Kelleher I & Cannon M (2010). Psychotic-like experiences in the general population: characterizing a high-risk group
for psychosis. Psychol Med. 19:1-6.
Kendell R & Jablensky A (2003). Distinguishing between the validity and utility of psychiatric diagnoses. Am J of
Psychiatry, 160 (1), 4-12.
Kendler KS, Heath AC, Martin NG, Eaves LJ (1987). Symptoms of anxiety and symptoms of depression. Same genes,
different environments? Arch Gen Psychiatry, 44(5):451-7.
Kendler KS, Neale MC, Kessler RC, Heath AC, Eaves LJ (1992). Major depression and generalized anxiety disorder.
Same genes, (partly) different environments? Arch Gen Psychiatry, 49(9):716-22.
160
Kendler KS (1996). Major depression and generalised anxiety disorder. Same genes, (partly)different environments-revisited. Br J Psychiatry, Suppl. (30):68-75.
Loonen AJ, Peer PG, Zwanikken GJ (1991). Continuation and maintenance therapy with antidepressive agents. Metaanalysis of research. Pharm Weekbl Sci, Aug 13(4):167-75.
Mortensen PB, Pedersen MG & Pedersen CB (2010). Psychiatric family history and schizophrenia risk in Denmark:
which mental disorders are relevant? Psychol Med, 40(2), 201-210.
Murray RM, Sham P, Van Os J, Zanelli J, Cannon M, McDonald C (2004). A developmental model for similarities and
dissimilarities between schizophrenia and bipolar disorder. Schizophr Res, 71(2-3):405-16.
Myin-Germeys I & Van Os J (2007). Stress-reactivity in psychosis: evidence for an affective pathway to psychosis. Clin
Psychol Rev, 27(4), 409-424.
Pantelis C, Yücel M, Wood SJ, McGorry PD, Velakoulis D (2003). Early and late neurodevelopmental disturbances in
schizophrenia and their functional consequences. Aust N Z J Psychiatry, 37(4):399-406.
Polanczyk G, Moffitt TE, Arseneault L, Cannon M, Ambler A, Keefe RS, Houts R, Odgers CL & Caspi A (2010). Etiological and clinical features of childhood psychotic symptoms: results from a birth cohort. Arch Gen Psychiatry,
67(4), 328-338.
Stip E, Letourneau G (2009). Psychotic symptoms as a continuum between normality and pathology. Can J Psychiatry, 54(3):140-51.
Van Os J, Hanssen M, Bijl R & Ravelli A (2000). Straus (1969) revisited: A Psychosis continuüm in the general population? Schizophrenia Res, 45 (1-2), 11-20.
Van Os J, Hanssen M, Bak M, Bijl RV, Vollebergh W (2003a). Do urbanicity and familial liability coparticipate in causing psychosis? Am J Psychiatry, 160(3):477-82.
Van Os J, Sham P, (2003b). Gene-environment interactions. In: Murray, R.M, Jones, P.B, Susser, E, et al. (Eds.), The
Epidemiology of Schizophrenia. Cambridge University Press, Cambridge, pp. 235–254.
Van Os J, Pedersen CB, Mortensen PB (2004). Confirmation of synergy between urbanicity and familial liability in the
causation of psychosis. Am J Psychiatry, 161(12):2312-4.
Van Os J & Kapur S (2009). Schizophrenia. Lancet, 374(9690):635-45.
Van Os J, Linscott RJ, Myin-Germeys I, Delespaul P & Krabbendam L (2009). A systematic review and meta-analysis of
the psychosis continuum: evidence for a psychosis proneness-persistence-impairment model of psychotic disorder. Psychol Med, 39(2), 179-195.
Van Os J & Rutten BPF (2009).Gene-Environment-Wide Interaction Studies in Psychiatry. Am J Psychiatry 166:964966.
Van Os J, Kenis G, Rutten BP (2010). The Environment and Schizophrenia; Clues for Genetic and Biological Discovery.
Nature, Nov 11;468(7321):203-12.
Van Os J (2010). Are psychiatric diagnosis of psychosis scientific and useful? The case of schizophrenia. Journal of
Mental Health 19 (4): 305-317.
Van Praag, HM (1976). About the impossible concept of schizophrenia. Compr Psychiatry, Jul-Aug; 17 (4), 481-497.
Viguera AC, Baldessarini RJ, Friedberg J (1998). Discontinuing antidepressant treatment in major depression. Harv
Rev Psychiatry, 5(6):293-306.
Weiser M, Reichenberg A, Kravitz E, Lubin G, Shmushkevich M, Glahn DC, Gross R, Rabinowitz J, Noy S & Davidson M
(2008). Subtle cognitive dysfunction in nonaffected siblings of individuals affected by non-psychotic disorders.
Biol Psychiatry, 63(6), 602-608.
Wiersma D, Nienhuis FJ, Slooff CJ, Giel R (1998). Natural course of schizophrenic disorders: a 15-year follow-up of a
Dutch incidence cohort. Schizophr Bull.;24(1):75-85.
161
Dankwoord
Het is erg moeilijk om één van de coryfeeën in psychiatrie en onderzoekswereld te bedanken... ik heb het over mijn promotor prof. dr. J. van Os. Waarmee moet ik beginnen... met
zijn bodemloze statistische en wetenschappelijke kennis? Met de vele wetenschappelijke
publicaties en jaren achtereen tot medisch specialist van het jaar benoemd te worden? Of
hoe hij stigmatisering in de psychiatrie, en de bijdrage van onze diagnostische systemen
daaraan, bespreekbaar maakt? Hij is een van de meest inspirerende hoogleraren voor de
GGZ professionals, hij brengt wetenschap dicht bij de clinici.
Jim, ik geloof dat ik één van je ‘enfants terribles’ (of misschien de enige) ben geweest,
maar ik dank je voor je betrokkenheid en uithoudingsvermogen dit proefschrift tot een goed
einde te brengen. Wat mij het meest opgevallen is, dat jij mij en mijn promotie nooit opgegeven hebt, ondanks het feit dat ik bij momenten door allerlei teleurstellingen niet verder
wilde..... en nog steeds vraag ik me af of ik het verdiend heb en of mij al deze eer toekomt.
Ik dank de leden van de leescommissie: prof. dr. M. de Vries, dr. Ph. Delespaul, dr. F. Peeters,
prof. dr. D. Wiersma en prof. dr. W.A. Nolen.
Prof. Nolen, beste Willem, ik wil je in het bijzonder bedanken, omdat ik van jou de kans
heb gekregen om aan onderzoek te proeven tijdens mijn arts-assistentschap en mijn eerste
artikel, een meta-analyse over onderhoudsbehandeling van depressie, met je te schrijven.
Ook ben je gaandeweg de promotie continu betrokken geweest en heb je met een aantal
artikelen mee gepubliceerd. Ik heb zeer veel van je geleerd: ik dank je hiervoor!
Prof. Wiersma wil ik bedanken voor zijn enthousiasme in het laatste traject van de promotie
om het tot een afronding te brengen.
Lex Wunderink, mijn eerste (ja, ik heb er vele gehad!) A-opleider: Lex, ik dank je voor de
kansen die je me gegeven hebt: te beginnen met de mogelijkheid om met de A-opleiding van
start te gaan! Van jou heb ik vooral geleerd om goed psychiatrisch onderzoek en goede diagnostiek te doen; altijd in beweging te blijven en jezelf te ontwikkelen... ook jij doet veel dingen tegelijkertijd en gaat op nieuwe uitdagingen af.
Een dankwoord richt ik ook graag tot Peter Turpijn, voorzitter van Mediant, alsmede Henk
van den Berg, lid van de Raad van Bestuur, voor de jarenlange ondersteuning en hulp bij het
tot stand brengen van mijn promotie en het faciliteren daarin. Peter, ik kan je niet genoeg
bedanken voor je eindeloze energieke en motiverende houding; Henk, ik wil je bedanken
voor je voortdurende ondersteuning, de inspiratie en de coaching die ik van je heb gekregen,
niet alleen als beginnend psychiater, maar ook in mijn huidige baan als manager behandelzaken, en daarnaast gedurende mijn hele promotietraject. Zelfs toen ik een jaar in de USA
zat, bleef je me adviezen geven en steunen bij alle perikelen rondom promotie. Ook was het
163
nodig dat je de laatste maanden tijdens elke managementvergadering zowat als vast agendapunt mijn promotietraject aan de orde liet komen, om me continu met de afronding ervan
bezig te houden, voor het geval dat ik teveel met andere zaken bezig was.
De samenwerking met jullie ervaar ik als zeer prettig en zeer dynamisch.
Dank ook aan de leden van het managementteam van Mediant voor hun ondersteuning
gedurende mijn promotietraject: Paul Deursen, manager Bedrijfsvoering, bij Acute en Curatieve Zorg, voor voor de fantastische samenwerking met veel humor en daadkracht! Willem
Snelleman, Hans Agelink en tot recent Hesie Chung, managers behandelzaken, Jurrien Smit,
Marcel Hooch Antink, Gabriëlle Parel, Vincent van Antwerpen, Frans van de Pol, Annelies ter
Huurne, Arie Holkers, Yvette Loves, Janet Poels, Ger Geuke, Wynand Koekkoek, allen lid van
het managementteam bij Mediant, dank ik voor hun samenwerking. Peter Dingemans wil ik
bedanken voor alle steun en begeleiding bij de laatste fase van mijn promotieonderzoek;
Peter, dank voor alle uitleg en de voorbereidingen voor mijn promotie!
Bovendien wil ik mijn collega’s bij Acute en Curatieve Zorg, de team- en programmanagers
bedanken voor het geduld dat ze hebben uitgeoefend ondanks alle drukte en de weinige tijd
die er was voor overleg of patiëntencontacten. Met name de collega’s van crisis-voordeurteam en bemoeizorg zijn zeer begripvol geweest in deze fase en hebben naast de dagelijkse
bezigheden steeds hun interesse getoond voor mijn promotieonderzoek. Er was afgelopen
jaar zelfs weinig tijd over om de krant te lezen, maar dankzij jullie was ik toch op de hoogte
van de interviews die mijn promotor had gegeven in allerlei dagbladen.
De secretaresses van het circuitmanagement en A&CZ wil ik bedanken, met name Bibiane, Sandra en Petra, voor al hun hulp en inzet in de laatste fase van mijn promotietraject, ik
was zeer blij met jullie ondersteuning. De secretaresses van het crisis-voordeurteam heb ik
soms, misschien wel -erg -vaak, tot wanhoop gedreven met voortdurende wijzigingen van
afspraken en aanpassingen in de agenda, maar zonder het eeuwige geduld van Jeanne, Juliette, Ingrid en Tineke was werken aan mijn promotie het afgelopen jaar niet mogelijk geweest.
Aan de UM heb ik zeer veel hulp gehad van Elsa Misdom met het realiseren van dit
proefschrift; Elsa, dank voor al je hulp!
In het bijzonder dank ik Ellemieke Nederlof, psychiater en vriendin, niet alleen voor alle steun
en aanmoediging voor mijn onderzoek, maar ook voor de leuke sociale contacten. Ellemieke,
ik wil je enorm bedanken voor je steun en hulp en ik ben zeer vereerd dat je mijn paranimf
wilt zijn. Ik hoop dat we nu wat rustiger tijden tegemoet gaan, al is het de vraag of wij in
staat zijn ons bezig te houden met maar één of twee dingen en niet met tig dingen tegelijkertijd. Ik dacht altijd dat ik na de promotie in een groot gat, ook wel ‘post-promotie depressie’
genoemd, zou vallen, omdat ik dan wellicht veel te veel tijd over zou houden. Maar, als ik zie
hoe druk jij nog steeds bent na je promotie, en hoe we ons continu nieuwe dingen op de hals
halen, dan maak ik me daar niet meer zo druk over... tegen die tijd heb ik niet eens meer tijd
om depressief te worden!
Chantal Theunissen wil ik bedanken voor de leuke en ontspannen momenten, waarbij we
over totaal andere zaken dan werk en onderzoek konden praten. Chantal, super, om je na
zoveel jaren weer terug te zien en dat je mijn paranimf wilt zijn; ik was gelukkig net op tijd,
164
om in je paranimf- agenda, die bijna volgeboekt is voor het voorjaar, nog een dag te kunnen
claimen voor mijn promotie!
Ik wil verder Fred Dreef en zijn vrouw Janine bedanken voor de gekkigheid, humor en gezellige gesprekken… dank Fred! Dankzij jouw humor kon ik dingen beter relativeren en soms
zodanig dat ik een paar maanden voor het afronden zowat alles opgaf om meer van het
leven te kunnen genieten... al duurde dat maar een paar seconden, het was wel geweldig!
Een zorgeloos bestaan, helemaal overtuigd van de zinloosheid van ons bestaan!
Lieve Serap, hoe vaak hebben we onze afspraken niet uitgesteld vanwege de zoveelste
deadline... het gaat binnenkort echt lukken om af te spreken! En geweldig, dat je temidden
van alle drukte, stress en slapeloze nachten mij opbelde, om me uit te nodigen voor een
trektocht door het Midden-Oosten en de zijderoute naar China! Misschien dit jaar dan?
Isik, jij was mijn eerste leermeester in de psychiatrie en hebt me, tijdens mijn coschappen, laten zien wat voor een geweldig vak psychiatrie is! Nog altijd leg ik mensen uit
dat ik niet de enige Koerdische psychiater in Nederland ben en dat jij de eerste was!
Verder wil ik alle vrienden die altijd betrokken, steunend zijn geweest bedanken voor
hun betrokkenheid; Diederik ten Back, Tekleh Zandi, Pieter de Nijs, Albert Meyer en natuurlijk Agnieszka Ederveen, dank voor alles!
Lieve Nicole, we gaan nu vaker uit!
… en nu even met z’n allen omschakelen naar een andere taal: Rozelin Aydina, benim Dersimli bilim arkadasima tesekkur etmek istiyorum... sen, Turkiyeden ta buralara geldin, bilimle
ugrasmak icin, oysa ben burada, bu rahatliklar icinde bazen sikayetleniyorum, Sana cok saygi
duyuyorum, tabii adinin Rozelin olmasida herseyi daha anlamli yapiyor; biliyorsun, benim
dunyalarca sevdegim yegenimin adida Rozelin!
Dear Prof. N. Freimer, I would like to thank you for giving me the opportunity to work at your
lab at the UCLA and learn the basics of science and be very critical, even about the biological
findings, which most researchers and clinicians seem to be accepting as conclusive. It was a
wonderful year and I did learn a lot during that year!
My dear Ania Jansinska, how can I thank you for the wonderful time we have spent together at the UCLA and in Los Angeles/California! I miss you a lot and miss our visits to botanical gardens, to the opera and into the nature! You’re probably the only scientist I know
who has two PhDs and I admire your relentless focus on good methodology, analyses and
also your modesty. Well, I hope we can meet soon in California or in your homeland Poland.
Dear Emery and JD; thanks for all the fun we had in LA! Hope to visit you guys soon!
Debra, querida amiga, muchas gracias por todo! Ojala, encontramos en verano en Sueca....of
all places!
En last but not least… wat zou ik zijn zonder mijn familie, en daarbij bedoel ik mijn grote
familie, dus niet alleen mijn broer, mijn twee zusjes en mijn moeder en vader, maar ook mijn
tantes, ooms, neefjes, nichtjes en oma. Bij Koerden is het niet gebruikelijk familie te bedanken voor het feit dat ze er voor je zijn, dat is zoiets waardevols en tegelijkertijd vanzelfsprekends, dat je dat niet eens ter discussie of ter sprake kunt brengen door ze te bedanken… het is ergens ook ‘not done’... en toch ga ik het wel een beetje doen.
165
Zoals dat hoort in een groot Koerdisch gezin is iedereen erg betrokken bij elke fase van je
leven en zo ook mijn familie. Ze hebben me gesteund vanaf het begin tot het eind. Al was het
soms moeilijk te begrijpen waar ik mee bezig was, jullie zijn er altijd voor mij geweest. Niet
alleen tijdens mijn promotietraject, maar vanaf het begin dat ik naar de universiteit ging tot
en met het afronden van het specialisme psychiatrie, tot het promotieonderzoek.
Ook de bezorgde momenten dat jullie af en toe me zelfs adviseerden om ‘mijn hersenen
rust te geven’.
Ercan dayi ve Fatos yengeye tesekkurler! Insallah, gelecek sefer Dersime geldigimde sizinle rahat rahat oturup oradaki doganin guzelliginin tadina varirim. Bu sefer bilgi sayarimi
getirmiyecegim!
Murat dayi, cher oncle, et Sidar, Berivan, Serkan et ma cherie Meral, Paris n’est pas très
loin, mais j’ai toujours manque du temps... mais après juin nous allons nous voir souvent, et
surtout à Paris, comme d’habitude!
Lieve Thea en Asmen, Yasar dayi is niet meer bij ons, maar hij is mijn grootste inspirator
geweest voor de psyche van de mens en zo ook van grote invloed op mijn keuze om psychiatrie te doen!
Mijn andere ooms en tantes en mijn oma wil ik bedanken voor al hun steun, ook mijn
neefjes en nichtjes en hun partners: Serda en Siska en Nino, Laser, Murat, Seval, Ozgur, Ozge,
Aynur teyze, Ayten teyze en Pierre, Unal dayi en Songul, met jullie geniet ik van de familiemomenten... na al die jaren is het nog steeds amazing om te zien met hoeveel mensen we
soms in een kleine woonkamer passen bij zulke gelegenheden... hoe doen we dat toch?
Mijn zusje Meral wil ik bedanken voor alle hulp die ik de afgelopen jaren heb gekregen
om dit promotieonderzoek af te kunnen ronden. Er is geen wereldstad waar we naar afgereisd zijn waar we het niet over mijn promotieperikelen hebben gehad. Je eeuwige steun
om vooruit te kijken, ondanks tegenslagen, is geweldig! Zonder jou had ik het waarschijnlijk
niet gered, je bent een grote inspiratiebron geweest.
Mijn zusje Safak dank ik voor de bemoedigende woorden, de rationaliteit waarmee je
naar de zaken kijkt en de nuchterheid waarmee je dingen aanhoorde en zorgvuldig je adviezen gaf. Ik heb er vijf jaar over gedaan om hier iets van te bakken en intussen ben jij in de
afgelopen jaren moeder geworden. Jij en Kurtulus, mijn zwager, geweldig dat jullie je dochtertje, mijn schatje Rozelin Sarya, regelmatig aan mij toevertrouwen in de weekenden (met
mijn moeder op de achtergrond weliswaar)... tja, heb weinig maternale instincten, maar ben
volgens mij wel een leuk maatje voor Rozelin! En als ik eens afdwaalde tijdens het spelen,
omdat ik aan werk of onderzoek dacht, dan kreeg ik een klap in mijn gezicht van Rozelin om
mijn aandacht weer bij het spelletje te houden! Kijk, dat zijn nou momenten waar je dankbaar voor bent, dat zo’n kleintje van 3 jaar je even terug in de realiteit brengt.
Mijn broer Akansel wil ik bedanken voor de rust en kalmte en laat ik het zeggen, toch de
structuur, die je steeds bij mij, je tegenpool, wist aan te brengen. Telkens als ik in paniek
raakte en het niet meer zag zitten, wist jij mij gerust te stellen en met je zakelijke kijk op
dingen tot logische keuzes te brengen... En als ik eens weer radicale beslissingen had genomen, dan was jij er altijd om snel wat ‘damage control’ te doen. Je bent er altijd voor mij
geweest en je bent een enorme steun voor me, maakt niet uit in welke stad ik woon, of in
welk land ik me op dat moment bevind of in welke fase van mijn leven. De tijden die ik met
jou, je vrouw Dagmar en de kinderen, lieve Miro, Amber en Sofie heb kunnen doorbrengen,
dat zijn geweldige momenten! Mijn vader wil ik bedanken voor de inspiratie die ik van hem
166
heb gekregen en voor de discipline. We hebben weinig tijd om elkaar te zien, maar van jou
heb ik veel geleerd, met name niet opgeven, risico’s durven nemen, vallen en opstaan en
gaan voor je ambities...
Mijn lieve moeder, waar moet ik beginnen... ergens denk ik dat jij vandaag ook promoveert, dat er ook een last van jou afvalt, omdat je de hoeveelheid stress die ik heb gehad
de afgelopen jaren net zo goed doorgemaakt hebt. Sevgili annecigim, Derman hanim, sensiz
hic bir yere kavusamazdim!
167
Curriculum vitae
Nil Kaymaz was born on 2 July 1972 in Mazgirt, East Turkey (Kurdish section). She completed
her pre-university education in the Netherlands at the Christelijke Streeklyceum in Ede,
Gelderland. After that, she studied medicine in Belgium, the first three years at the Free
University of Brussel and the last four years at Ghent University. She did part of her clinical
rotations at the Vall d’Hebron University Hospital in Barcelona, Spain. In 1999, Nil completed
her medical studies and began working as a junior doctor in the adolescent clinic at the
Sophia Children's Hospital in Rotterdam. Subsequently, she worked as a junior doctor at
Midden-Brabant Mental Healthcare (GGZ) agency in Tilburg, the Netherlands. In 2000, she
worked as a junior doctor at Grote Rivieren, an organization for mental health care in
Dordrecht, the Netherlands, to begin her specialization in psychiatry. She completed her
education in psychiatry in 2005 and then went to work as a psychiatrist/team psychiatrist at
Altrecht Mental Healthcare agency in Utrecht. In the same year, she was awarded a Mozaïek
grant from The Netherlands Organization for Scientific Research (NWO) and she began her
postdoctoral research. Since 2005, Nil has been following the Master's programme in Epidemiology given by the Free University of Amsterdam/Institute for Health and Care Research
(EMGO). In 2008, she carried out research on neuroimaging for one year at the University of
California in Los Angeles (UCLA) in the United States, where she also lectured college students in psychopathology and neuroimaging. In 2009, after returning to the Netherlands, she
started working at Arkin in Amsterdam as a psychiatrist/psychiatrist administrator in longterm care.
During the winter of 2009-2010, she moved to a post at the Mediant Mental Healthcare
agency in Enschede, the Netherlands, where she is currently employed as treatment manager of the curative division and as psychiatrist with the critical time intervention team.
169
Publications
1. Kaymaz N, Lataster T, Lieb R, Wittchen H-U, Van Os J, Drukker M. The case of the missing
evidence: what do subclinical psychosis spectrum experiences predict in unselected representative population samples? A systematic review enriched with new results. Submitted, 2010.
2. Kaymaz N, Van Os J. DSM-5 and the 'Psychosis Risk Syndrome': Babylonic confusion.
Psychosis, Volume 2, Issue 2 June 2010, pages 100 - 103
3. Kaymaz N, Van Os J. Extended psychosis phenotype – yes: single continuum – unlikely.
Psychol Med. 2010 Mar 18:1-4.
4. Kaymaz N, Van Os J. Heritability of structural brain traits: an endophenotype approach to
deconstruct schizophrenia. Int Rev Neurobiol. 2009;89:85-130.
5. Kaymaz N, Van Os J. Murray et al. (2004) revisited: is bipolar disorder identical to schizophrenia without developmental impairment? Acta Psychiatr Scand. 2009 Oct;120(4):24952.
6. Kaymaz N, Van Os J, Loonen AJ, Nolen WA. Evidence that patients with single versus
recurrent depressive episodes are differentially sensitive to treatment discontinuation: a
meta-analysis of placebo-controlled randomized trials. J Clin Psychiatry. 2008
Sep;69(9):1423-36.
7. Kaymaz N, Van Os J, de Graaf R, Ten Have M, Nolen W, Krabbendam L. The impact of
subclinical psychosis on the transition from subclinical mania to bipolar disorder. J Affect
Disord. 2007 Feb;98(1-2):55-64
8. Kaymaz N, Krabbendam L, de Graaf R, Nolen W, Ten Have M, Van Os J. Evidence that the
urban environment specifically impacts on the psychotic but not the affective dimension
of bipolar disorder. Soc Psychiatry Psychiatr Epidemiol. 2006 Sep;41(9):679-85.
171